Feature Engieering with different ML models.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature Extraction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Combining Chroma and mfcc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Processing Negative-con\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 7\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Processing Postive-con\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 2\n",
      "(12,)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Furqan Ali\\Anaconda3\\envs\\tensorflow_env\\lib\\site-packages\\librosa\\core\\pitch.py:145: UserWarning: Trying to estimate tuning from empty frequency set.\n",
      "  warnings.warn('Trying to estimate tuning from empty frequency set.')\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 7\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 0\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 1\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 2\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 3\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 4\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 5\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 6\n",
      "(12,)\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 7\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "DATASET_PATH = \"cough_cleaned_touseef\"\n",
    "JSON_PATH = \"new_features.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    # 1) Loop through all the genres\n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment  # if s=0 -> num_segments_per_sample\n",
    "                    chroma = librosa.feature.chroma_stft(y=signal[start_sample: finish_sample], sr=sr,n_chroma=12, n_fft=4096)\n",
    "                    cq = librosa.feature.chroma_cqt(y=chroma[0], sr=sr)\n",
    "                    gh = cq.reshape(cq.shape[0]* cq.shape[1])\n",
    "                    print(gh.shape)\n",
    "                    \n",
    "                    mfcc = librosa.feature.mfcc(gh,\n",
    "                                                sr =sr,\n",
    "                                                n_fft = n_fft,\n",
    "                                                n_mfcc = n_mfcc,\n",
    "                                                hop_length = hop_length\n",
    "                                                )\n",
    "#                     mfcc = mfcc.T\n",
    "                    \n",
    "#                     if len(mfcc) == expected_no_of_mfcc_vectors_per_segment:\n",
    "    #                     mfcc = mfcc.T   # making mfcc vector suitable to our DL model.\n",
    "                    data[\"mfcc\"].append(mfcc.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Combining mfcc and zero-crossing rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Processing Negative-con\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 7\n",
      "\n",
      " Processing Postive-con\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 7\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 7\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "\n",
    "\n",
    "DATASET_PATH = \"cough_cleaned_touseef\"\n",
    "JSON_PATH = \"mfcc_zerocross.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    # 1) Loop through all the genres\n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment  # if s=0 -> num_segments_per_sample\n",
    "#                     chroma = librosa.feature.chroma_stft(y=signal[start_sample: finish_sample], sr=sr,n_chroma=12, n_fft=4096)\n",
    "#                     cq = librosa.feature.chroma_cqt(y=chroma[0], sr=sr)\n",
    "#                     gh = cq.reshape(cq.shape[0]* cq.shape[1])\n",
    "#                     print(gh.shape)\n",
    "                    \n",
    "                    mfcc = librosa.feature.mfcc(signal[start_sample: finish_sample],\n",
    "                                                sr =sr,\n",
    "                                                n_fft = n_fft,\n",
    "                                                n_mfcc = n_mfcc,\n",
    "                                                hop_length = hop_length\n",
    "                                                )\n",
    "                    \n",
    "                    hhh = mfcc.reshape(mfcc.shape[0]*mfcc.shape[1])\n",
    "\n",
    "                    zcr = librosa.feature.zero_crossing_rate(hhh)\n",
    "                    zcr = zcr.T\n",
    "                    \n",
    "#                     if len(mfcc) == expected_no_of_mfcc_vectors_per_segment:\n",
    "    #                     mfcc = mfcc.T   # making mfcc vector suitable to our DL model.\n",
    "                    data[\"mfcc\"].append(zcr.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# USING Roll off Features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Processing Negative-con\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 7\n",
      "\n",
      " Processing Postive-con\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 7\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 7\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "\n",
    "\n",
    "DATASET_PATH = \"cough_cleaned_touseef\"\n",
    "JSON_PATH = \"roll_off.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment \n",
    "                    \n",
    "                    rolloff = librosa.feature.spectral_rolloff(y=signal[start_sample: finish_sample], sr=sr, roll_percent=0.1)\n",
    "#                     chroma = librosa.feature.chroma_stft(y=signal[start_sample: finish_sample], sr=sr,n_chroma=12, n_fft=4096)\n",
    "#                     cq = librosa.feature.chroma_cqt(y=chroma[0], sr=sr)\n",
    "#                     gh = cq.reshape(cq.shape[0]* cq.shape[1])\n",
    "#                     print(gh.shape)\n",
    "                    \n",
    "#                     mfcc = librosa.feature.mfcc(signal[start_sample: finish_sample],\n",
    "#                                                 sr =sr,\n",
    "#                                                 n_fft = n_fft,\n",
    "#                                                 n_mfcc = n_mfcc,\n",
    "#                                                 hop_length = hop_length\n",
    "#                                                 )\n",
    "                    \n",
    "#                     hhh = mfcc.reshape(mfcc.shape[0]*mfcc.shape[1])\n",
    "\n",
    "#                     zcr = librosa.feature.zero_crossing_rate(hhh)\n",
    "                    rolloff = rolloff.T\n",
    "                    \n",
    "#                     if len(mfcc) == expected_no_of_mfcc_vectors_per_segment:\n",
    "    #                     mfcc = mfcc.T   # making mfcc vector suitable to our DL model.\n",
    "                    data[\"mfcc\"].append(rolloff.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Combining any three features:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 311,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Processing Negative-con\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 7\n",
      "\n",
      " Processing Postive-con\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 7\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 7\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import numpy as np\n",
    "import json\n",
    "\n",
    "\n",
    "DATASET_PATH = \"cough_cleaned_touseef\"\n",
    "JSON_PATH = \"rolloff_rms.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment \n",
    "                    \n",
    "                    rms = librosa.feature.rms(y = signal[start_sample: finish_sample])\n",
    "                    rms_sha = rms.reshape(rms.shape[0]*rms.shape[1])\n",
    "                    rolloff = librosa.feature.spectral_rolloff(y=rms_sha, sr=sr, roll_percent=0.1)\n",
    "                    rolloff_sha = rolloff.reshape(rolloff.shape[0]*rolloff.shape[1])\n",
    "                    S = np.abs(librosa.stft(rolloff_sha))\n",
    "                    p2 = librosa.feature.poly_features(S=S, order=2)\n",
    "                    \n",
    "                    \n",
    "                    data[\"mfcc\"].append(p2.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Combining 3 imp features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 319,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Processing Negative-con\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 7\n",
      "\n",
      " Processing Postive-con\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 7\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 7\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "\n",
    "\n",
    "DATASET_PATH = \"cough_cleaned_touseef\"\n",
    "JSON_PATH = \"rms_roll_mfcc.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment \n",
    "                    \n",
    "                    rms = librosa.feature.rms(y = signal[start_sample: finish_sample])\n",
    "                    rms_sha = rms.reshape(rms.shape[0]*rms.shape[1])\n",
    "                    rolloff = librosa.feature.spectral_rolloff(y=rms_sha, sr=sr, roll_percent=0.1)\n",
    "                    rolloff_sha = rolloff.reshape(rolloff.shape[0]*rolloff.shape[1])\n",
    "                    \n",
    "                    mfcc = librosa.feature.mfcc(rolloff_sha,\n",
    "                                                sr =sr,\n",
    "                                                n_fft = n_fft,\n",
    "                                                n_mfcc = n_mfcc,\n",
    "                                                hop_length = hop_length\n",
    "                                                )\n",
    "                    \n",
    "#                     mfcc = mfcc.T\n",
    "                    \n",
    "#                      if len(mfcc) == expected_no_of_mfcc_vectors_per_segment:\n",
    "    #                     mfcc = mfcc.T   # making mfcc vector suitable to our DL model.\n",
    "                    data[\"mfcc\"].append(mfcc.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# USING Roll oFF and MFCC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 357,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      " Processing Negative-con\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-17.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-20.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-21.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Negative-con\\neg-9.wav, segment : 7\n",
      "\n",
      " Processing Postive-con\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-1.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-10.wav, segment : 7\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-11.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-12.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-13.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-14.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-15.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-16.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-2.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-3.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-4.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-5.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-6.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-7.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-8.wav, segment : 7\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 0\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 1\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 2\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 3\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 4\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 5\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 6\n",
      "cough_cleaned_touseef\\Postive-con\\pos-9.wav, segment : 7\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "\n",
    "\n",
    "DATASET_PATH = \"cough_cleaned_touseef\"\n",
    "JSON_PATH = \"roll_mfcc.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment \n",
    "                    \n",
    "                    rolloff = librosa.feature.spectral_rolloff(y=signal[start_sample: finish_sample], sr=sr, roll_percent=0.1)\n",
    "                    rolloff_sha = rolloff.reshape(rolloff.shape[0]*rolloff.shape[1])\n",
    "                    \n",
    "                    mfcc = librosa.feature.mfcc(rolloff_sha,\n",
    "                                                sr =sr,\n",
    "                                                n_fft = n_fft,\n",
    "                                                n_mfcc = n_mfcc,\n",
    "                                                hop_length = hop_length\n",
    "                                                )\n",
    "                    \n",
    "#                     mfcc = mfcc.T\n",
    "                    \n",
    "#                     if len(mfcc) == expected_no_of_mfcc_vectors_per_segment:\n",
    "#                         mfcc = mfcc.T   # making mfcc vector suitable to our DL model.\n",
    "                    data[\"mfcc\"].append(mfcc.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Using Rolloff and MFCC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "\n",
    "DATASET_PATH = \"recent_toufeeq_cough_data\"\n",
    "JSON_PATH = \"roll_mfcc_testing1.json\"\n",
    "SAMPLE_RATE = 22050\n",
    "DURATION = 8  # measured in seconds.\n",
    "SAMPLES_PER_TRACK = SAMPLE_RATE * DURATION\n",
    "\n",
    "\n",
    "def save_mfcc(dataset_path, json_path, n_mfcc = 13, n_fft = 2048, hop_length= 512, num_segments = 5):\n",
    "   \n",
    "    # dictionary to store data\n",
    "    data = {\n",
    "        \"mapping\": [],\n",
    "        \"mfcc\": [],\n",
    "        \"labels\": []\n",
    "    }\n",
    "    \n",
    "    num_samples_per_segment =  int(SAMPLES_PER_TRACK / num_segments)\n",
    "    \n",
    "    expected_no_of_mfcc_vectors_per_segment = math.ceil(num_samples_per_segment / hop_length) \n",
    "    \n",
    "    for i, (dirpath, dirname, filenames) in enumerate(os.walk(dataset_path)):\n",
    "        \n",
    "        # Ensure that we r not at the root level\n",
    "        if dirpath is not dataset_path:\n",
    "            dirpath_components = dirpath.split(\"\\\\\")  # \"genre/blues\" => [\"genre\", \"blues\"]\n",
    "            semantic_label = dirpath_components[-1]\n",
    "            data[\"mapping\"].append(semantic_label)\n",
    "            \n",
    "            print(\"\\n Processing {}\".format(semantic_label))\n",
    "            \n",
    "            # Process files for specific genre. \n",
    "            for f in filenames:\n",
    "                \n",
    "                # laoding audio file\n",
    "                file_path = os.path.join(dirpath, f)\n",
    "                signal, sr = librosa.load(file_path, sr = SAMPLE_RATE)\n",
    "                \n",
    "                # divide signals into segments,process segments, extract mfcc and at last store mfcc vectors.\n",
    "                for s in range(num_segments):\n",
    "                    start_sample = num_samples_per_segment * s  # if s= 0 -> then start_sample = 0 \n",
    "                    finish_sample = start_sample + num_samples_per_segment \n",
    "                    \n",
    "                    rolloff = librosa.feature.spectral_rolloff(y=signal[start_sample: finish_sample], sr=sr, roll_percent=0.1)\n",
    "                    rolloff_sha = rolloff.reshape(rolloff.shape[0]*rolloff.shape[1])\n",
    "                    \n",
    "                    mfcc = librosa.feature.mfcc(rolloff_sha,\n",
    "                                                sr =sr,\n",
    "                                                n_fft = n_fft,\n",
    "                                                n_mfcc = n_mfcc,\n",
    "                                                hop_length = hop_length\n",
    "                                                )\n",
    "                    \n",
    "#                     mfcc = mfcc.T\n",
    "                    \n",
    "#                     if len(mfcc) == expected_no_of_mfcc_vectors_per_segment:\n",
    "#                         mfcc = mfcc.T   # making mfcc vector suitable to our DL model.\n",
    "                    data[\"mfcc\"].append(mfcc.tolist()) # we can not append mfcc directly blc its a numpy array so we have\n",
    "                    # to first convert it into list.\n",
    "\n",
    "                    data[\"labels\"].append(i-1)\n",
    "                    # i mean in each iteration we are in different genre folder. why r we doing i -1 that's blc in first \n",
    "                    # iteration we were not inside the dir containing all the genres folder. \n",
    "\n",
    "                    print(\"{}, segment : {}\".format(file_path, s))\n",
    "\n",
    "    # final step: saving everything as a json file            \n",
    "    with open(json_path, \"w\") as fp:\n",
    "        json.dump(data, fp, indent = 4)  # indent mean spaces while writing. fp mean file_path  \n",
    "        \n",
    "        \n",
    "if __name__ == \"__main__\":\n",
    "    save_mfcc(DATASET_PATH, JSON_PATH, num_segments = 8)\n",
    "                "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# For visualizing "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(13, 454)\n",
      "(12, 454)\n",
      "(1025, 454)\n",
      "(1, 454)\n",
      "(454, 1051)\n"
     ]
    }
   ],
   "source": [
    "import os \n",
    "import librosa\n",
    "import math\n",
    "import json\n",
    "import librosa.display\n",
    "\n",
    "path = \"/Users/Furqan Ali/Desktop/saylani_python/Coursera_learning/virufy_cleaned_file/cough_cleaned_touseef/Postive-con/pos-1.wav\"\n",
    "sr = 22050\n",
    "\n",
    "\n",
    "# Zero crossing rate\n",
    "y, sr = librosa.load(path, sr = sr)\n",
    "zc = librosa.feature.chroma_stft(y=y, sr=sr,n_chroma=12, n_fft=4096)\n",
    "\n",
    "mfcc = librosa.feature.mfcc(y,\n",
    "                                sr =sr,\n",
    "                                n_fft = 2048,\n",
    "                                n_mfcc = 13,\n",
    "                                hop_length = 512\n",
    "                                )\n",
    "                    \n",
    "chroma_cq = librosa.feature.chroma_cqt(y=y, sr=sr)\n",
    "\n",
    "pitches, magnitudes = librosa.piptrack(y=y, sr=sr)\n",
    "rolloff = librosa.feature.spectral_rolloff(y=y, sr=sr, roll_percent=0.1)\n",
    "\n",
    "features = np.concatenate((mfcc, chroma_cq, pitches, rolloff), axis = 0)\n",
    "\n",
    "features = features.T\n",
    "# zc = librosa.feature.zero_crossing_rate(y)\n",
    "\n",
    "# print(spectral_entropy(zc, 100, method='welch', normalize=True))\n",
    "\n",
    "# librosa.display.waveplot(zc[0], sr)\n",
    "\n",
    "print(mfcc.shape)\n",
    "print(chroma_cq.shape)\n",
    "print(pitches.shape)\n",
    "print(rolloff.shape)\n",
    "print(features.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for plotting chroma graph\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "plt.figure(figsize=(10, 4))\n",
    "librosa.display.specshow(features, y_axis='chroma', x_axis='time')\n",
    "plt.colorbar()\n",
    "plt.title('Chromagram')\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# so here is very imp note for all \n",
    "# why dont we just make many segments of our audio sample and then classiy all those segments al together and then take a mean\n",
    "# of all those samples and give 1 output so lets say if we have noise in the cough so normally when we have multiple segments \n",
    "# so some of segents will be containing noise and some of them will be containng meaningfull iformation and so when u do \n",
    "# classification on all those segments so most prbbaly the part which is conating cough will always predict right and the one \n",
    "# containng nouse will not get predicted well. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 485,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(256, 13, 1)\n",
      "(13, 1)\n"
     ]
    }
   ],
   "source": [
    "# Random trial for testing stored data\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "print(X.shape)\n",
    "print(X[0].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM model 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.495\n",
      "The accuracy score on test set is : 0.49514563106796117\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"chroma.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=10)\n",
    "\n",
    "# training = X_train.reshape((X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "# training = np.reshape(X_train, (X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C=20, gamma=0.00001)\n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.515\n",
      "The accuracy score on test set is : 0.5145631067961165\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"chroma_mfcc.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=51)\n",
    "\n",
    "# training = X_train.reshape((X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "# training = np.reshape(X_train, (X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C=20, gamma=0.00001)\n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 220,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.495\n",
      "The accuracy score on test set is : 0.49514563106796117\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"mfcc_zerocross.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=31)\n",
    "\n",
    "# training = X_train.reshape((X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "# training = np.reshape(X_train, (X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C=20, gamma=0.00001)\n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM with rolloff features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 235,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.655\n",
      "The accuracy score on test set is : 0.6551724137931034\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"roll_off.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.45, random_state=30)\n",
    "\n",
    "# training = X_train.reshape((X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "# training = np.reshape(X_train, (X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C=20, gamma=0.00001)\n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Roll off is performing better on SVM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM with 3 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 309,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.495\n",
      "The accuracy score on test set is : 0.49514563106796117\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"3_combined_features.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=35)\n",
    "\n",
    "# training = X_train.reshape((X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "# training = np.reshape(X_train, (X_train.shape[1], X_train.shape[2]))\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C=20, gamma=0.01)\n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM 3 famous combined features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 484,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.476\n",
      "The accuracy score on test set is : 0.47572815533980584\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"rms_roll_mfcc.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=17)\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C=3, gamma=0.0001)\n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM ROLL+MFCC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 467,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "fitting...\n",
      "acc=0.660\n",
      "The accuracy score on test set is : 0.6601941747572816\n"
     ]
    }
   ],
   "source": [
    "# Without normalising\n",
    "\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=30)\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "# training = X_train.shape[1], X_train.shape[2])\n",
    "# Simple SVM\n",
    "print('fitting...')\n",
    "clf = SVC(C = 3, gamma = 0.0001)      \n",
    "# min no of c at which accuracy doesnt decrease\n",
    "clf.fit(training, y_train)\n",
    "acc = clf.score(test, y_test)\n",
    "print(\"acc=%0.3f\" % acc)\n",
    "predicted=clf.predict(test)\n",
    "print(\"The accuracy score on test set is :\",  accuracy_score(y_test,predicted))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random_forest \n",
    "\n",
    "## Roll+mfcc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 496,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model is running .....\n",
      "Testing Accuracy: 0.6504854368932039\n",
      "\n",
      "ROC curve value is:  0.6982237339380196\n",
      "Confusion matrix, without normalization\n",
      "[[33 16]\n",
      " [20 34]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "\n",
    "def load_data(dataset_path):\n",
    "    with open(dataset_path, \"r\") as fp:\n",
    "        data = json.load(fp)\n",
    "        \n",
    "    # converting lists into numpy arrays\n",
    "    X = np.array(data[\"mfcc\"])\n",
    "    y = np.array(data[\"labels\"])\n",
    "    \n",
    "    return X, y\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "# normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "# Split data into training and test subsets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=45, shuffle=True)\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "#Create a Gaussian Classifier\n",
    "clf=RandomForestClassifier(max_depth=50, random_state=48, n_estimators=97)\n",
    "\n",
    "#Train the model using the training sets y_pred=clf.predict(X_test)\n",
    "clf.fit(training, y_train)\n",
    "\n",
    "y_pred=clf.predict(test)\n",
    "\n",
    "# Model Accuracy, how often is the classifier correct\n",
    "\n",
    "print(\"Model is running .....\")\n",
    "print(\"Testing Accuracy:\", metrics.accuracy_score(y_test, y_pred))\n",
    "\n",
    "# Probabilities for each class\n",
    "rf_probs = clf.predict_proba(test)[:, 1]\n",
    "\n",
    "# Calculate roc auc\n",
    "roc_value = roc_auc_score(y_test, rf_probs)\n",
    "print()\n",
    "print(\"ROC curve value is: \",roc_value)\n",
    "\n",
    "\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Oranges):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    Source: http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html\n",
    "    \"\"\"\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    print(cm)\n",
    "\n",
    "    # Plot the confusion matrix\n",
    "    plt.figure(figsize = (10, 10))\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(title, size = 24)\n",
    "    plt.colorbar(aspect=4)\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45, size = 14)\n",
    "    plt.yticks(tick_marks, classes, size = 14)\n",
    "\n",
    "    fmt = '.2f' if normalize else 'd'\n",
    "    thresh = cm.max() / 2.\n",
    "    \n",
    "    # Labeling the plot\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, format(cm[i, j], fmt), fontsize = 20,\n",
    "                 horizontalalignment=\"center\",\n",
    "                 color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "        \n",
    "    plt.grid(None)\n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label', size = 18)\n",
    "    plt.xlabel('Predicted label', size = 18)\n",
    "\n",
    "# Confusion matrix\n",
    "cm = confusion_matrix(y_test, y_pred)\n",
    "plot_confusion_matrix(cm, classes = ['Covid', 'Not Covid'],\n",
    "                      title = 'Health Confusion Matrix')\n",
    "\n",
    "plt.savefig('cm78.png')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Making random forest class"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools\n",
    "\n",
    "\n",
    "class RF_Model:\n",
    "    def __init__(self, dataset_path, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators):\n",
    "        self.dataset_path = dataset_path\n",
    "        self.test_size = test_size\n",
    "        self.ran_state = ran_state\n",
    "        self.rf_maxdepth = rf_maxdepth\n",
    "        self.rf_ran_state = rf_ran_state\n",
    "        self.rf_estimators = rf_estimators\n",
    "    \n",
    "\n",
    "    def load_data(self):\n",
    "        with open(self.dataset_path, \"r\") as fp:\n",
    "            data = json.load(fp)\n",
    "\n",
    "        # converting lists into numpy arrays\n",
    "        X = np.array(data[\"mfcc\"])\n",
    "        y = np.array(data[\"labels\"])\n",
    "\n",
    "        return X, y\n",
    "    \n",
    "    def model(self):\n",
    "        X, y = self.load_data()\n",
    "\n",
    "        # normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "        # Split data into training and test subsets\n",
    "        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=self.test_size, random_state=self.ran_state, \n",
    "                                                            shuffle=True)\n",
    "\n",
    "        training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "        test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "        accuracy_history = []\n",
    "        \n",
    "        for max_depth in range(1, self.rf_maxdepth):\n",
    "            for estimators in range(1, rf_estimators):\n",
    "                #Create a Random forest Clcassifier\n",
    "                clf=RandomForestClassifier(max_depth=max_depth, random_state=self.rf_ran_state,\n",
    "                                           n_estimators = estimators)\n",
    "\n",
    "                #Train the model using the training sets y_pred=clf.predict(X_test)\n",
    "                clf.fit(training, y_train)\n",
    "\n",
    "                y_pred=clf.predict(test)\n",
    "                accuracy = accuracy_score(y_pred,y_test)\n",
    "                print(\"Accuracy on test dataset with maxDepth: \", max_depth, \"estimators\", estimators,\"Accuracy: \", accuracy)\n",
    "\n",
    "        # Model Accuracy, how often is the classifier correct\n",
    "\n",
    "        print(\"Model is running .....\")\n",
    "        print(\"Testing Accuracy:\", metrics.accuracy_score(y_test, y_pred))\n",
    "\n",
    "        # Probabilities for each class\n",
    "        rf_probs = clf.predict_proba(test)[:, 1]\n",
    "\n",
    "        # Calculate roc auc\n",
    "        roc_value = roc_auc_score(y_test, rf_probs)\n",
    "        print()\n",
    "        print(\"ROC curve value is: \",roc_value)\n",
    "        print()\n",
    "        # Confusion matrix\n",
    "        cm = confusion_matrix(y_test, y_pred)\n",
    "        self.plot_confusion_matrix(cm, classes = ['Covid', 'Not Covid'],\n",
    "                              title = 'Health Confusion Matrix')\n",
    "\n",
    "        plt.savefig('dadasdsd.png')\n",
    "        \n",
    "        return (clf, y_pred)\n",
    "\n",
    "    def plot_confusion_matrix(self, cm, classes,\n",
    "                              normalize=False,\n",
    "                              title='Confusion matrix',\n",
    "                              cmap=plt.cm.Oranges):\n",
    "        \"\"\"\n",
    "        This function prints and plots the confusion matrix.\n",
    "        Normalization can be applied by setting `normalize=True`.\n",
    "        Source: http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html\n",
    "        \"\"\"\n",
    "        if normalize:\n",
    "            cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "            print(\"Normalized confusion matrix\")\n",
    "        else:\n",
    "            print('Confusion matrix, without normalization')\n",
    "\n",
    "        print(cm)\n",
    "\n",
    "        # Plot the confusion matrix\n",
    "        plt.figure(figsize = (10, 10))\n",
    "        plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "        plt.title(title, size = 24)\n",
    "        plt.colorbar(aspect=4)\n",
    "        tick_marks = np.arange(len(classes))\n",
    "        plt.xticks(tick_marks, classes, rotation=45, size = 14)\n",
    "        plt.yticks(tick_marks, classes, size = 14)\n",
    "\n",
    "        fmt = '.2f' if normalize else 'd'\n",
    "        thresh = cm.max() / 2.\n",
    "\n",
    "        # Labeling the plot\n",
    "        for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "            plt.text(j, i, format(cm[i, j], fmt), fontsize = 20,\n",
    "                     horizontalalignment=\"center\",\n",
    "                     color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "        plt.grid(None)\n",
    "        plt.tight_layout()\n",
    "        plt.ylabel('True label', size = 18)\n",
    "        plt.xlabel('Predicted label', size = 18)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "ff = [i for i in range(600)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  1 estimators 1 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 2 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 3 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 4 Accuracy:  0.3333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 5 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 6 Accuracy:  0.3333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 7 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 8 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 9 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 10 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 11 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 12 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 13 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 14 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 15 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 16 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 17 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 18 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 19 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 20 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 21 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 22 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 23 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 24 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 25 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 26 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 27 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 28 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 29 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 30 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 31 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 32 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 33 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 34 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 35 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 36 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 37 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 38 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 39 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 40 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 41 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 42 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 43 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 44 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 45 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 46 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 47 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 48 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 49 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 50 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 51 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 52 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 53 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 54 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 55 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 56 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 57 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 58 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 59 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 60 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 61 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 62 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 63 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 64 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 65 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 66 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 67 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 68 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 69 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 70 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 71 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 72 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 73 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 74 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 75 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 76 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 77 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 78 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 79 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 80 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 81 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 82 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 83 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 84 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 85 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 86 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  1 estimators 87 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 1 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 2 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 3 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 4 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 5 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 6 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 7 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 8 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 9 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 10 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 11 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 12 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 13 Accuracy:  0.4666666666666667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  2 estimators 14 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 15 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 16 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 17 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 18 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 19 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 20 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 21 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 22 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 23 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 24 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 25 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 26 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 27 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 28 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 29 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 30 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 31 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 32 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 33 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 34 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 35 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 36 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 37 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 38 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 39 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 40 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 41 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 42 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 43 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 44 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 45 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 46 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 47 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 48 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 49 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 50 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 51 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 52 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 53 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 54 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 55 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 56 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 57 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 58 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 59 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 60 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 61 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 62 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 63 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 64 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 65 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 66 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 67 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 68 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 69 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 70 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 71 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 72 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 73 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 74 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 75 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 76 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 77 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 78 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 79 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 80 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 81 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 82 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 83 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 84 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 85 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 86 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  2 estimators 87 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 1 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 2 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 3 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 4 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 5 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 6 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 7 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 8 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 9 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 10 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 11 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 12 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 13 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 14 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 15 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 16 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 17 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 18 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 19 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 20 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 21 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 22 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 23 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 24 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 25 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 26 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 27 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 28 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 29 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 30 Accuracy:  0.4\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  3 estimators 31 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 32 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 33 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 34 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 35 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 36 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 37 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 38 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 39 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 40 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 41 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 42 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 43 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 44 Accuracy:  0.3333333333333333\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 45 Accuracy:  0.3333333333333333\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 46 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 47 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 48 Accuracy:  0.36666666666666664\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 49 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 50 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 51 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 52 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 53 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 54 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 55 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 56 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 57 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 58 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 59 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 60 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 61 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 62 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 63 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 64 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 65 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 66 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 67 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 68 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 69 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 70 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 71 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 72 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 73 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 74 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 75 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 76 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 77 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 78 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 79 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 80 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 81 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 82 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 83 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 84 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 85 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 86 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  3 estimators 87 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 1 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 2 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 3 Accuracy:  0.6666666666666666\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 4 Accuracy:  0.6666666666666666\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 5 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 6 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 7 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 8 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 9 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 10 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 11 Accuracy:  0.6666666666666666\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 12 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 13 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 14 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 15 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 16 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 17 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 18 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 19 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 20 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 21 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 22 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 23 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 24 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 25 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 26 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 27 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 28 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 29 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 30 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 31 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 32 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 33 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 34 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 35 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 36 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 37 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 38 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 39 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 40 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 41 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 42 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 43 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 44 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 45 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 46 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 47 Accuracy:  0.5666666666666667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  4 estimators 48 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 49 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 50 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 51 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 52 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 53 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 54 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 55 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 56 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 57 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 58 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 59 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 60 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 61 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 62 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 63 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 64 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 65 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 66 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 67 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 68 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 69 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 70 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 71 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 72 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 73 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 74 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 75 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 76 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 77 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 78 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 79 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 80 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 81 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 82 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 83 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 84 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 85 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 86 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  4 estimators 87 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 1 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 2 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 3 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 4 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 5 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 6 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 7 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 8 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 9 Accuracy:  0.7\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 10 Accuracy:  0.7\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 11 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 12 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 13 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 14 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 15 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 16 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 17 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 18 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 19 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 20 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 21 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 22 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 23 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 24 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 25 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 26 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 27 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 28 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 29 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 30 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 31 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 32 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 33 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 34 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 35 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 36 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 37 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 38 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 39 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 40 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 41 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 42 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 43 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 44 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 45 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 46 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 47 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 48 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 49 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 50 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 51 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 52 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 53 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 54 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 55 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 56 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 57 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 58 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 59 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 60 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 61 Accuracy:  0.4666666666666667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  5 estimators 62 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 63 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 64 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 65 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 66 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 67 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 68 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 69 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 70 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 71 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 72 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 73 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 74 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 75 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 76 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 77 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 78 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 79 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 80 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 81 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 82 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 83 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 84 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 85 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 86 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  5 estimators 87 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 1 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 2 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 3 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 4 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 5 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 6 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 7 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 8 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 9 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 10 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 11 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 12 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 13 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 14 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 15 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 16 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 17 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 18 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 19 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 20 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 21 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 22 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 23 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 24 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 25 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 26 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 27 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 28 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 29 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 30 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 31 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 32 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 33 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 34 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 35 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 36 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 37 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 38 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 39 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 40 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 41 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 42 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 43 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 44 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 45 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 46 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 47 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 48 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 49 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 50 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 51 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 52 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 53 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 54 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 55 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 56 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 57 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 58 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 59 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 60 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 61 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 62 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 63 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 64 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 65 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 66 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 67 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 68 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 69 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 70 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 71 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 72 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 73 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 74 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 75 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 76 Accuracy:  0.5666666666666667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  6 estimators 77 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 78 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 79 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 80 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 81 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 82 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 83 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 84 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 85 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 86 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  6 estimators 87 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 1 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 2 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 3 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 4 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 5 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 6 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 7 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 8 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 9 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 10 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 11 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 12 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 13 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 14 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 15 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 16 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 17 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 18 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 19 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 20 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 21 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 22 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 23 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 24 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 25 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 26 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 27 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 28 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 29 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 30 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 31 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 32 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 33 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 34 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 35 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 36 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 37 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 38 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 39 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 40 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 41 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 42 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 43 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 44 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 45 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 46 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 47 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 48 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 49 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 50 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 51 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 52 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 53 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 54 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 55 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 56 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 57 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 58 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 59 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 60 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 61 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 62 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 63 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 64 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 65 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 66 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 67 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 68 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 69 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 70 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 71 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 72 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 73 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 74 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 75 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 76 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 77 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 78 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 79 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 80 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 81 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 82 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 83 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 84 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 85 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 86 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  7 estimators 87 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 1 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 2 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 3 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 4 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 5 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 6 Accuracy:  0.5666666666666667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  8 estimators 7 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 8 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 9 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 10 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 11 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 12 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 13 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 14 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 15 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 16 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 17 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 18 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 19 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 20 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 21 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 22 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 23 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 24 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 25 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 26 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 27 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 28 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 29 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 30 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 31 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 32 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 33 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 34 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 35 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 36 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 37 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 38 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 39 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 40 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 41 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 42 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 43 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 44 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 45 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 46 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 47 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 48 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 49 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 50 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 51 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 52 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 53 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 54 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 55 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 56 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 57 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 58 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 59 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 60 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 61 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 62 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 63 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 64 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 65 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 66 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 67 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 68 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 69 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 70 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 71 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 72 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 73 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 74 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 75 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 76 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 77 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 78 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 79 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 80 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 81 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 82 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 83 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 84 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 85 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 86 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  8 estimators 87 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 1 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 2 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 3 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 4 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 5 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 6 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 7 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 8 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 9 Accuracy:  0.6333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 10 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 11 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 12 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 13 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 14 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 15 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 16 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 17 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 18 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 19 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 20 Accuracy:  0.4666666666666667\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on test dataset with maxDepth:  9 estimators 21 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 22 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 23 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 24 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 25 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 26 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 27 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 28 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 29 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 30 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 31 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 32 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 33 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 34 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 35 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 36 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 37 Accuracy:  0.4\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 38 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 39 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 40 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 41 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 42 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 43 Accuracy:  0.43333333333333335\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 44 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 45 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 46 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 47 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 48 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 49 Accuracy:  0.4666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 50 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 51 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 52 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 53 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 54 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 55 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 56 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 57 Accuracy:  0.5\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 58 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 59 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 60 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 61 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 62 Accuracy:  0.6\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 63 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 64 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 65 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 66 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 67 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 68 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 69 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 70 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 71 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 72 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 73 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 74 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 75 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 76 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 77 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 78 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 79 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 80 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 81 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 82 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 83 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 84 Accuracy:  0.5666666666666667\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 85 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 86 Accuracy:  0.5333333333333333\n",
      "Accuracy on test dataset with maxDepth:  9 estimators 87 Accuracy:  0.5333333333333333\n",
      "Model is running .....\n",
      "Testing Accuracy: 0.5333333333333333\n",
      "\n",
      "ROC curve value is:  0.5833333333333333\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[10  8]\n",
      " [ 6  6]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calling random forest class here:\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc_test.json\"\n",
    "\n",
    "test_size = 30\n",
    "ran_state = 12\n",
    "rf_maxdepth = 10\n",
    "rf_ran_state = 171\n",
    "rf_estimators = 88\n",
    "\n",
    "\n",
    "accuracy_history = []\n",
    "\n",
    "\n",
    "rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "# rf.load_data\n",
    "clf, y_pred = rf.model()\n",
    "#         print(\"Accuracy on test dataset with\", max_depth, \"kernel\", estimators,\"degree\",k,\"gamma\",l,\"C: \", accuracy)\n",
    "#                 accuracy_history.append(accuracy)\n",
    "#         accuracy_history.append(accuracy)\n",
    "        \n",
    "# test_size = 30\n",
    "# ran_state = 12\n",
    "# rf_maxdepth = 10\n",
    "# rf_ran_state = 166\n",
    "# rf_estimators = 88\n",
    "\n",
    "# rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "# # rf.load_data\n",
    "# rf.model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model is running .....\n",
      "Testing Accuracy: 0.8\n",
      "\n",
      "ROC curve value is:  0.8312020460358056\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[13  4]\n",
      " [ 4 19]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calling random forest class here with different parameters:\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "test_size = 40\n",
    "ran_state = 27\n",
    "rf_maxdepth = 10\n",
    "rf_ran_state = 171\n",
    "rf_estimators = 89\n",
    "\n",
    "rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "\n",
    "rf.model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 648,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model is running .....\n",
      "Testing Accuracy: 0.8\n",
      "\n",
      "ROC curve value is:  0.8312020460358056\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[13  4]\n",
      " [ 4 19]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calling random forest class here with different parameters:\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "test_size = 40\n",
    "ran_state = 27\n",
    "rf_maxdepth = 10\n",
    "rf_ran_state = 171\n",
    "rf_estimators = 89\n",
    "\n",
    "rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "\n",
    "rf.model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 929,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.96      0.97      0.96       312\n",
      "           1       0.95      0.94      0.94       216\n",
      "\n",
      "    accuracy                           0.95       528\n",
      "   macro avg       0.95      0.95      0.95       528\n",
      "weighted avg       0.95      0.95      0.95       528\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test,y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# RANDOM FOREST MODEL WITH CROSS VALIDATION:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 688,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import numpy as np\n",
    "import sklearn\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import KFold\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn import preprocessing\n",
    "import json\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn import metrics\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import itertools\n",
    "from sklearn.metrics import classification_report\n",
    "        \n",
    "\n",
    "\n",
    "class RF_Model:\n",
    "    def __init__(self, dataset_path, n_splits, ran_state, rf_maxdepth, rf_ran_state, rf_estimators):\n",
    "        self.dataset_path = dataset_path\n",
    "        self.n_splits = n_splits\n",
    "        self.ran_state = ran_state\n",
    "        self.rf_maxdepth = rf_maxdepth\n",
    "        self.rf_ran_state = rf_ran_state\n",
    "        self.rf_estimators = rf_estimators\n",
    "    \n",
    "\n",
    "    def load_data(self):\n",
    "        with open(self.dataset_path, \"r\") as fp:\n",
    "            data = json.load(fp)\n",
    "\n",
    "        # converting lists into numpy arrays\n",
    "        X = np.array(data[\"mfcc\"])\n",
    "        y = np.array(data[\"labels\"])\n",
    "\n",
    "        return X, y\n",
    "    \n",
    "    def model(self):\n",
    "        X, y = load_data(self.dataset_path)\n",
    "\n",
    "        # normalized_X = preprocessing.normalize(X)\n",
    "\n",
    "        # Split data \n",
    "        kf = KFold(n_splits=self.n_splits, random_state=self.ran_state, shuffle=True) # Define the split - into 2 folds\n",
    "        \n",
    "        for train_index, test_index in kf.split(X):\n",
    "            X_train, X_test = X[train_index], X[test_index]\n",
    "            y_train, y_test = y[train_index], y[test_index]\n",
    "        \n",
    "#         X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=self.test_size, random_state=self.ran_state, \n",
    "#                                                             shuffle=True)\n",
    "\n",
    "        training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "        test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "\n",
    "        #Create a Random forest Classifier\n",
    "        clf=RandomForestClassifier(max_depth=self.rf_maxdepth, random_state=self.rf_ran_state,\n",
    "                                   n_estimators = self.rf_estimators)\n",
    "\n",
    "        #Train the model using the training sets y_pred=clf.predict(X_test)\n",
    "        clf.fit(training, y_train)\n",
    "\n",
    "        y_pred=clf.predict(test)\n",
    "\n",
    "        # Model Accuracy, how often is the classifier correct\n",
    "\n",
    "        print(\"Model is running .....\")\n",
    "        print(\"Testing Accuracy:\", metrics.accuracy_score(y_test, y_pred))\n",
    "\n",
    "        # Probabilities for each class\n",
    "        rf_probs = clf.predict_proba(test)[:, 1]\n",
    "\n",
    "        # Calculate roc auc\n",
    "        roc_value = roc_auc_score(y_test, rf_probs)\n",
    "        print()\n",
    "        print(\"ROC curve value is: \",roc_value)\n",
    "        print()\n",
    "        \n",
    "        print(classification_report(y_test,y_pred))\n",
    "        \n",
    "        # Confusion matrix\n",
    "        cm = confusion_matrix(y_test, y_pred)\n",
    "        plot_confusion_matrix(cm, classes = ['Covid', 'Not Covid'],\n",
    "                              title = 'Health Confusion Matrix')\n",
    "\n",
    "        plt.savefig('dasd.png')\n",
    "        \n",
    "        return clf\n",
    "    \n",
    "    def plot_confusion_matrix(cm, classes,\n",
    "                              normalize=False,\n",
    "                              title='Confusion matrix',\n",
    "                              cmap=plt.cm.Oranges):\n",
    "        \"\"\"\n",
    "        This function prints and plots the confusion matrix.\n",
    "        Normalization can be applied by setting `normalize=True`.\n",
    "        Source: http://scikit-learn.org/stable/auto_examples/model_selection/plot_confusion_matrix.html\n",
    "        \"\"\"\n",
    "        if normalize:\n",
    "            cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "            print(\"Normalized confusion matrix\")\n",
    "        else:\n",
    "            print('Confusion matrix, without normalization')\n",
    "\n",
    "        print(cm)\n",
    "\n",
    "        # Plot the confusion matrix\n",
    "        plt.figure(figsize = (10, 10))\n",
    "        plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "        plt.title(title, size = 24)\n",
    "        plt.colorbar(aspect=4)\n",
    "        tick_marks = np.arange(len(classes))\n",
    "        plt.xticks(tick_marks, classes, rotation=45, size = 14)\n",
    "        plt.yticks(tick_marks, classes, size = 14)\n",
    "\n",
    "        fmt = '.2f' if normalize else 'd'\n",
    "        thresh = cm.max() / 2.\n",
    "\n",
    "        # Labeling the plot\n",
    "        for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "            plt.text(j, i, format(cm[i, j], fmt), fontsize = 20,\n",
    "                     horizontalalignment=\"center\",\n",
    "                     color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "\n",
    "        plt.grid(None)\n",
    "        plt.tight_layout()\n",
    "        plt.ylabel('True label', size = 18)\n",
    "        plt.xlabel('Predicted label', size = 18)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 692,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model is running .....\n",
      "Testing Accuracy: 1.0\n",
      "\n",
      "ROC curve value is:  1.0\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00         4\n",
      "           1       1.00      1.00      1.00         2\n",
      "\n",
      "    accuracy                           1.00         6\n",
      "   macro avg       1.00      1.00      1.00         6\n",
      "weighted avg       1.00      1.00      1.00         6\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[4 0]\n",
      " [0 2]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calling random forest class here with k-fold cross validation:\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "n_splits = 2\n",
    "ran_state = None\n",
    "rf_maxdepth = 10\n",
    "rf_ran_state = 171\n",
    "rf_estimators = 89\n",
    "\n",
    "rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "\n",
    "model = rf.model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# So the above accuracy is obtained on 32 samples, now down below I m gonna test this model on 110 sample points. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# HERE TESTING YOUR Testing data seperately one by one"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 748,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1 1 0 0 0 1 1 0 1 1 0 0 1 1 1 1 1 0 0 1 0 0 1 1 0 0 1 0 0 1 0 0 0 0 1 0\n",
      " 1 1 1 1 0 1 1 0 1 0 0 0 1 0 0 1 0 0 0 0 0 1 0 0 0 1 0 1 0 1 0 0 0 1 0 0 0\n",
      " 1 0 1 0 1 1 1 1 0 0 0 1 0 1 0 1 1 1 1 1 1 0 0 1 1 1 1 1 1 0 0 0 0 0 1 1 0\n",
      " 1 0 1 0 1 1 0 1 1 1 0 0 0 1 0 1 0 1 1 0 1 1 0 0 0 0 0 1 1 1 1 0 1 1 1 1 1\n",
      " 1 0 1 1 0 0 0 1 0 1 1 1 1 1 0 0 1 1 0 0 0 0 1 0 0 0 1 1 0 1 1 1 1 0 1 0 1\n",
      " 1 0 0 0 1 0 1 1 1 0 1 0 0 1 0 1 1 1 0 1 0 0 0 1 0 0 1 0 0 0 0 1 0 0 1 0 1\n",
      " 0 0 1 1 1 1 1 0 1 0 1 1 0 0 0 0 0 0 1 1 0 0 1 1 0 0 1 1 1 0 1 0 0 1 0 1 0\n",
      " 0 1 1 1 0 1 1 1 1 0 0 0 0 1 0 1 0 0 1 0 0 1 0 1 0 0 0 0 0 0 1 0 1 0 0 0 0\n",
      " 0 0 1 0 0 0 0 1 1 0 1 0 1 1 0 0 1 0 0 0 0 0 0 1 1 0 0 0 0 1 1 1 1 0 0 1 0\n",
      " 0 1 0 0 0 1 1 1 0 1 1 0 1 1 0 0 0 0 0]\n",
      "Testing Accuracy: 0.6875\n"
     ]
    }
   ],
   "source": [
    "DATASET_PATH = \"roll_mfcc_testing1.json\"\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.40, random_state=49, shuffle=True)\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "y_pred = model.predict(test)\n",
    "print(y_pred)\n",
    "print(\"Testing Accuracy:\", metrics.accuracy_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training random forest on new dataset containing 110 samples\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 928,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model is running .....\n",
      "Testing Accuracy: 0.9090909090909091\n",
      "\n",
      "ROC curve value is:  0.8470588235294118\n",
      "\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.89      1.00      0.94        17\n",
      "           1       1.00      0.60      0.75         5\n",
      "\n",
      "    accuracy                           0.91        22\n",
      "   macro avg       0.95      0.80      0.85        22\n",
      "weighted avg       0.92      0.91      0.90        22\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[17  0]\n",
      " [ 2  3]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# here i am using 5 folds.\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc_testing1.json\"\n",
    "n_splits = 5\n",
    "ran_state = None\n",
    "rf_maxdepth = 15\n",
    "rf_ran_state = 159\n",
    "rf_estimators = 89\n",
    "\n",
    "rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "\n",
    "model2 = rf.model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 870,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1 0 0 0 1 1 1 0 1 1 0 0 1 0 0 1 0 1 0 1 0 0 0 1 0 1 0 0 0 0 0 0 0 1 1 0\n",
      " 1 1 1 0 0 0 1 0 0 0 0 0 1 1 0 1 1 1 1 0 0 0 0 0 0 0 1 1 0 1 0 0 1 0 0 0 0\n",
      " 0 0 1 0 0 1 1 0 1 0 0 0 1 0 0 1 0 0 0 0 1 0 0 1 0 0 1 0 0 0 1 0 0 0 1 1 0\n",
      " 0 0 1 0 1 1 0 0 0 0 0 0 1 1 0 1 0 0 1 0 1 0 0 0 1 0 1 0 1 1 0 0 1 0 1 1 1\n",
      " 0 0 1 1 1 1 0 0 1 1 1 1 1 1 0 0 0 1 0 0 0 0 1 1 0 0 1 1 0 1 0 1 0 0 0 0 1\n",
      " 1 0 0 0 1 0 1 1 1 1 0 0 0 1 0 1 1 0 1 0 0 0 1 1 0 0 1 1 1 0 0 1 1 0 1 1 1\n",
      " 0 0 1 0 0 1 0 1 1 0 1 1 0 0 0 0 1 0 1 1 0 1 1 1 0 0 0 1 1 1 0 0 0 0 0 0 0\n",
      " 0 0 1 1 0 0 0 0 1 0 0 0 0 1 0 1 1 0 1 1 1 1 0 1 0 1 1 0 1 1 0 0 1 1 0 0 0\n",
      " 0 1 1 0 0 0 0 0 1 1 1 1 1 1 0 0 1 0 0 0 0 0 0 1 1 1 0 0 0 1 0 0 1 0 0 1 0\n",
      " 0 0 1 0 0 0 1 1 0 1 1 0 0 0 1 0 0 0 0 0 0 1 0 1 1 1 1 1 0 0 0 0 0 0 1 0 0\n",
      " 0 0 0 1 0 1 0 0 0 1 0 1 1 0 0 0 0 0 0 1 0 0 0 0 0 0 1 1 1 0 0 1 0 0 1 0 0\n",
      " 1 0 1 1 0 1 1 1 0 0 1 0 0 1 0 0 0 1 0 1 1 1 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0\n",
      " 0 0 1 1 0 1 0 0 1 1 0 0 0 1 0 1 1 0 1 0 0 0 0 0 0 1 1 0 0 0 1 1 1 1 1 0 0\n",
      " 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 1 1 1 0 0 1 1 0 0 1 1 0 0 0 0 1 0 0 1 1 0 0\n",
      " 1 0 0 0 1 1 0 1 0 1]\n",
      "Testing Accuracy: 0.9545454545454546\n"
     ]
    }
   ],
   "source": [
    "# Testing the model with test dataset.\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc_testing1.json\"\n",
    "\n",
    "X, y = load_data(DATASET_PATH)\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.60, random_state=49, shuffle=True)\n",
    "\n",
    "training = X_train.reshape(X_train.shape[0], (X_train.shape[1]*X_train.shape[2]))\n",
    "test = X_test.reshape(X_test.shape[0], (X_test.shape[1]*X_test.shape[2]))\n",
    "\n",
    "y_pred = model.predict(test)\n",
    "print(y_pred)\n",
    "print(\"Testing Accuracy:\", metrics.accuracy_score(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 671,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model is running .....\n",
      "Testing Accuracy: 1.0\n",
      "\n",
      "ROC curve value is:  1.0\n",
      "\n",
      "Confusion matrix, without normalization\n",
      "[[2 0]\n",
      " [0 4]]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Calling random forest class here with k-fold cross validation:\n",
    "\n",
    "DATASET_PATH = \"roll_mfcc.json\"\n",
    "n_splits = 5\n",
    "ran_state = None\n",
    "rf_maxdepth = 10\n",
    "rf_ran_state = 171\n",
    "rf_estimators = 89\n",
    "\n",
    "rf = RF_Model(DATASET_PATH, test_size, ran_state, rf_maxdepth, rf_ran_state, rf_estimators)\n",
    "\n",
    "model2 = rf.model()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For checking how much each feature is participating in the generalization procedure.\n",
    "\n",
    "df1 = pd.DataFrame.from_records(x_train)     \n",
    "tmp = pd.DataFrame({'Feature': df1.columns, 'Feature importance': clf_rf.feature_importances_})\n",
    "tmp = tmp.sort_values(by='Feature importance',ascending=False)\n",
    "plt.figure(figsize = (7,4))\n",
    "plt.title('Features importance',fontsize=14)\n",
    "s = sns.barplot(x='Feature',y='Feature importance',data=tmp)\n",
    "s.set_xticklabels(s.get_xticklabels(),rotation=90)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tensorflow_env",
   "language": "python",
   "name": "tensorflow_env"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}