Prosup_DTM_Binary.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ProSup DTM Binary Movement Classifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "from Data_preparation_Library import *\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tqdm.notebook import tqdm\n",
    "from time import time\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.models import load_model\n",
    "from random import choice\n",
    "from scipy import signal\n",
    "from pickle import load\n",
    "from sklearn.preprocessing import OneHotEncoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def filter_df(files_df,order=1,cf=50,fs=2000):\n",
    "    for i in range(len(files_df)):\n",
    "        b,a = signal.butter(1, 50,fs=fs)\n",
    "        emg_df = pd.DataFrame(columns=emg_labels)\n",
    "        for labels in emg_labels:\n",
    "            emg_df[labels] = signal.lfilter(b, a,files_df[i][labels])\n",
    "        files_df[i][emg_labels] = np.array(emg_df)\n",
    "        \n",
    "        b,a = signal.butter(3, 3,fs=2000)\n",
    "        for labels in angle_lables:\n",
    "            files_df[i][labels] = signal.lfilter(b, a,files_df[i][labels])\n",
    "    return files_df\n",
    "emg_labels = ['EMG1','EMG2','EMG3','EMG4','EMG5','EMG6','EMG7','EMG8']\n",
    "angle_lables = ['Pronation_Angle','Flexion_Angle',  'Radial_Angle','Elbow_Joint_Angle']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generate File Names"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pronation Supination Files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "prosup_files = []\n",
    "dtm_files = []\n",
    "subjects = ['Fazil','Nikhil','Ingy','Abhishek','Jai','Deepti','Nirmayee','Varun'] #'Fazil/day_04_14_21' \n",
    "# subjects = ['Varun'] #'Fazil/day_04_14_21' 'Varun'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Subjects/Fazil/computed_Bulb.csv',\n",
       " 'Subjects/Fazil/computed_Bulb01.csv',\n",
       " 'Subjects/Fazil/computed_Bulb02.csv',\n",
       " 'Subjects/Fazil/computed_Screw.csv',\n",
       " 'Subjects/Fazil/computed_Screw01.csv',\n",
       " 'Subjects/Fazil/computed_Screw02.csv',\n",
       " 'Subjects/Nikhil/computed_Bulb.csv',\n",
       " 'Subjects/Nikhil/computed_Bulb01.csv',\n",
       " 'Subjects/Nikhil/computed_Bulb02.csv',\n",
       " 'Subjects/Nikhil/computed_Screw.csv',\n",
       " 'Subjects/Nikhil/computed_Screw01.csv',\n",
       " 'Subjects/Nikhil/computed_Screw02.csv',\n",
       " 'Subjects/Ingy/computed_Bulb.csv',\n",
       " 'Subjects/Ingy/computed_Bulb01.csv',\n",
       " 'Subjects/Ingy/computed_Bulb02.csv',\n",
       " 'Subjects/Ingy/computed_Screw.csv',\n",
       " 'Subjects/Ingy/computed_Screw01.csv',\n",
       " 'Subjects/Ingy/computed_Screw02.csv',\n",
       " 'Subjects/Abhishek/computed_Bulb.csv',\n",
       " 'Subjects/Abhishek/computed_Bulb01.csv',\n",
       " 'Subjects/Abhishek/computed_Bulb02.csv',\n",
       " 'Subjects/Abhishek/computed_Screw.csv',\n",
       " 'Subjects/Abhishek/computed_Screw01.csv',\n",
       " 'Subjects/Abhishek/computed_Screw02.csv',\n",
       " 'Subjects/Jai/computed_Bulb.csv',\n",
       " 'Subjects/Jai/computed_Bulb01.csv',\n",
       " 'Subjects/Jai/computed_Bulb02.csv',\n",
       " 'Subjects/Jai/computed_Screw.csv',\n",
       " 'Subjects/Jai/computed_Screw01.csv',\n",
       " 'Subjects/Jai/computed_Screw02.csv',\n",
       " 'Subjects/Deepti/computed_Bulb.csv',\n",
       " 'Subjects/Deepti/computed_Bulb01.csv',\n",
       " 'Subjects/Deepti/computed_Bulb02.csv',\n",
       " 'Subjects/Deepti/computed_Screw.csv',\n",
       " 'Subjects/Deepti/computed_Screw01.csv',\n",
       " 'Subjects/Deepti/computed_Screw02.csv',\n",
       " 'Subjects/Nirmayee/computed_Bulb.csv',\n",
       " 'Subjects/Nirmayee/computed_Bulb01.csv',\n",
       " 'Subjects/Nirmayee/computed_Bulb02.csv',\n",
       " 'Subjects/Nirmayee/computed_Screw.csv',\n",
       " 'Subjects/Nirmayee/computed_Screw01.csv',\n",
       " 'Subjects/Nirmayee/computed_Screw02.csv',\n",
       " 'Subjects/Varun/computed_Bulb.csv',\n",
       " 'Subjects/Varun/computed_Bulb01.csv',\n",
       " 'Subjects/Varun/computed_Bulb02.csv',\n",
       " 'Subjects/Varun/computed_Screw.csv',\n",
       " 'Subjects/Varun/computed_Screw01.csv',\n",
       " 'Subjects/Varun/computed_Screw02.csv']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for subj in subjects:\n",
    "    path = 'Subjects/'+subj\n",
    "    # trial_names = ['Bulb','Hammer'] # 'Bulb','Hammer' ,'Cup','Screw'\n",
    "    trial_names = ['Bulb','Screw'] #,\n",
    "    for exp in trial_names:\n",
    "        for i in range(3):\n",
    "            if(i==0):\n",
    "                prosup_files.append(path+'/computed_'+exp+'.csv')\n",
    "            else:\n",
    "                prosup_files.append(path+'/computed_'+exp+'0'+str(i)+'.csv')\n",
    "prosup_files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "prosup_files_df = []\n",
    "for file in prosup_files:\n",
    "    prosup_files_df.append(pd.read_csv(file,index_col=0)[6500:85000]) #[4500:85000]\n",
    "prosup_files_df = filter_df(prosup_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Remove Test reps Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_rep_files = ['Subjects/Nikhil/computed_Bulb02.csv', \n",
    "                  'Subjects/Jai/computed_Bulb02.csv',\n",
    "                  'Subjects/Deepti/computed_Bulb02.csv'] #'Subjects/Deepti/computed_Bulb02.csv'\n",
    "tr_range = [72500,82500]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "prosup_files_df = []\n",
    "for file in prosup_files:\n",
    "    if file in test_rep_files:\n",
    "        df = pd.read_csv(file,index_col=0)\n",
    "        df = df.drop(df.index[tr_range[0]:tr_range[1]])\n",
    "    else:\n",
    "        df = pd.read_csv(file,index_col=0)[6500:85000]\n",
    "    prosup_files_df.append(df)\n",
    "prosup_files_df = filter_df(prosup_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Only Removed Reps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-2-ad34c0b128b2>:7: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  files_df[i][emg_labels] = np.array(emg_df)\n",
      "C:\\Users\\moham\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\pandas\\core\\indexing.py:1736: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  isetter(loc, value[:, i].tolist())\n",
      "<ipython-input-2-ad34c0b128b2>:11: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  files_df[i][labels] = signal.lfilter(b, a,files_df[i][labels])\n"
     ]
    }
   ],
   "source": [
    "prosup_files_df = []\n",
    "prosup_files = test_rep_files\n",
    "for file in prosup_files:\n",
    "    df = pd.read_csv(file,index_col=0)\n",
    "    prosup_files_df.append(df[tr_range[0]:tr_range[1]])\n",
    "prosup_files_df = filter_df(prosup_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Rearrange "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[['EMG8', 'EMG1', 'EMG2', 'EMG3', 'EMG4', 'EMG5', 'EMG6', 'EMG7'],\n",
       " ['EMG2', 'EMG3', 'EMG4', 'EMG5', 'EMG6', 'EMG7', 'EMG8', 'EMG1']]"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from collections import deque\n",
    "rearr_labels = []\n",
    "labels = deque(emg_labels)\n",
    "labels.rotate(1)\n",
    "rearr_labels.append(list(labels))\n",
    "labels.rotate(-2)\n",
    "rearr_labels.append(list(labels))\n",
    "rearr_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7e8515b3ed3b4bdaa95acf7140b23b5b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=3.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "9"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "files_df_rearrange = []\n",
    "for df in tqdm(prosup_files_df):\n",
    "    for lb in rearr_labels:\n",
    "        df[emg_labels] = df[lb]\n",
    "        files_df_rearrange.append(df) \n",
    "    \n",
    "prosup_files_df = prosup_files_df+files_df_rearrange\n",
    "len(prosup_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### DTM Files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Subjects/Fazil/computed_Hammer.csv',\n",
       " 'Subjects/Fazil/computed_Hammer01.csv',\n",
       " 'Subjects/Fazil/computed_Hammer02.csv',\n",
       " 'Subjects/Fazil/computed_Cup.csv',\n",
       " 'Subjects/Fazil/computed_Cup01.csv',\n",
       " 'Subjects/Fazil/computed_Cup02.csv',\n",
       " 'Subjects/Nikhil/computed_Hammer.csv',\n",
       " 'Subjects/Nikhil/computed_Hammer01.csv',\n",
       " 'Subjects/Nikhil/computed_Hammer02.csv',\n",
       " 'Subjects/Nikhil/computed_Cup.csv',\n",
       " 'Subjects/Nikhil/computed_Cup01.csv',\n",
       " 'Subjects/Nikhil/computed_Cup02.csv',\n",
       " 'Subjects/Ingy/computed_Hammer.csv',\n",
       " 'Subjects/Ingy/computed_Hammer01.csv',\n",
       " 'Subjects/Ingy/computed_Hammer02.csv',\n",
       " 'Subjects/Ingy/computed_Cup.csv',\n",
       " 'Subjects/Ingy/computed_Cup01.csv',\n",
       " 'Subjects/Ingy/computed_Cup02.csv',\n",
       " 'Subjects/Abhishek/computed_Hammer.csv',\n",
       " 'Subjects/Abhishek/computed_Hammer01.csv',\n",
       " 'Subjects/Abhishek/computed_Hammer02.csv',\n",
       " 'Subjects/Abhishek/computed_Cup.csv',\n",
       " 'Subjects/Abhishek/computed_Cup01.csv',\n",
       " 'Subjects/Abhishek/computed_Cup02.csv',\n",
       " 'Subjects/Jai/computed_Hammer.csv',\n",
       " 'Subjects/Jai/computed_Hammer01.csv',\n",
       " 'Subjects/Jai/computed_Hammer02.csv',\n",
       " 'Subjects/Jai/computed_Cup.csv',\n",
       " 'Subjects/Jai/computed_Cup01.csv',\n",
       " 'Subjects/Jai/computed_Cup02.csv',\n",
       " 'Subjects/Deepti/computed_Hammer.csv',\n",
       " 'Subjects/Deepti/computed_Hammer01.csv',\n",
       " 'Subjects/Deepti/computed_Hammer02.csv',\n",
       " 'Subjects/Deepti/computed_Cup.csv',\n",
       " 'Subjects/Deepti/computed_Cup01.csv',\n",
       " 'Subjects/Deepti/computed_Cup02.csv',\n",
       " 'Subjects/Nirmayee/computed_Hammer.csv',\n",
       " 'Subjects/Nirmayee/computed_Hammer01.csv',\n",
       " 'Subjects/Nirmayee/computed_Hammer02.csv',\n",
       " 'Subjects/Nirmayee/computed_Cup.csv',\n",
       " 'Subjects/Nirmayee/computed_Cup01.csv',\n",
       " 'Subjects/Nirmayee/computed_Cup02.csv',\n",
       " 'Subjects/Varun/computed_Hammer.csv',\n",
       " 'Subjects/Varun/computed_Hammer01.csv',\n",
       " 'Subjects/Varun/computed_Hammer02.csv',\n",
       " 'Subjects/Varun/computed_Cup.csv',\n",
       " 'Subjects/Varun/computed_Cup01.csv',\n",
       " 'Subjects/Varun/computed_Cup02.csv']"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for subj in subjects:\n",
    "    path = 'Subjects/'+subj\n",
    "    # trial_names = ['Bulb','Hammer'] # 'Bulb','Hammer' ,'Cup','Screw'\n",
    "    trial_names = ['Hammer','Cup'] #,\n",
    "    for exp in trial_names:\n",
    "        for i in range(3):\n",
    "            if(i==0):\n",
    "                dtm_files.append(path+'/computed_'+exp+'.csv')\n",
    "            else:\n",
    "                dtm_files.append(path+'/computed_'+exp+'0'+str(i)+'.csv')\n",
    "dtm_files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "dtm_files_df = []\n",
    "for file in dtm_files:\n",
    "    dtm_files_df.append(pd.read_csv(file,index_col=0)) #[9000:48000]\n",
    "dtm_files_df = filter_df(dtm_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Remove Test Rep Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_rep_files = ['Subjects/Nikhil/computed_Hammer02.csv', \n",
    "                  'Subjects/Jai/computed_Hammer02.csv',\n",
    "                 'Subjects/Deepti/computed_Hammer02.csv'] #'Subjects/Deepti/computed_Hammer02.csv'\n",
    "tr_range = [40000,50000]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "dtm_files_df = []\n",
    "for file in dtm_files:\n",
    "    if file in test_rep_files:\n",
    "        df = pd.read_csv(file,index_col=0)\n",
    "        df = df.drop(df.index[tr_range[0]:tr_range[1]])\n",
    "    else:\n",
    "        df = pd.read_csv(file,index_col=0)[9000:48000]\n",
    "    dtm_files_df.append(df)\n",
    "dtm_files_df = filter_df(dtm_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Only Removed Reps"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-2-ad34c0b128b2>:7: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  files_df[i][emg_labels] = np.array(emg_df)\n",
      "<ipython-input-2-ad34c0b128b2>:11: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  files_df[i][labels] = signal.lfilter(b, a,files_df[i][labels])\n"
     ]
    }
   ],
   "source": [
    "dtm_files_df = []\n",
    "dtm_files = test_rep_files\n",
    "for file in dtm_files:\n",
    "    df = pd.read_csv(file,index_col=0)\n",
    "    dtm_files_df.append(df[tr_range[0]:tr_range[1]])\n",
    "dtm_files_df = filter_df(dtm_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Rearrange "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[['EMG8', 'EMG1', 'EMG2', 'EMG3', 'EMG4', 'EMG5', 'EMG6', 'EMG7'],\n",
       " ['EMG2', 'EMG3', 'EMG4', 'EMG5', 'EMG6', 'EMG7', 'EMG8', 'EMG1']]"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from collections import deque\n",
    "rearr_labels = []\n",
    "labels = deque(emg_labels)\n",
    "labels.rotate(1)\n",
    "rearr_labels.append(list(labels))\n",
    "labels.rotate(-2)\n",
    "rearr_labels.append(list(labels))\n",
    "rearr_labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7b1c4045be6142908af2adca94051be4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=3.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "9"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "files_df_rearrange = []\n",
    "for df in tqdm(dtm_files_df):\n",
    "    for lb in rearr_labels:\n",
    "        df[emg_labels] = df[lb]\n",
    "        files_df_rearrange.append(df) \n",
    "    \n",
    "dtm_files_df = dtm_files_df+files_df_rearrange\n",
    "len(dtm_files_df)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Normalization Scaler Generate"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset Preparation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "window = 250\n",
    "interval = 20\n",
    "Normalize = True\n",
    "rms = False\n",
    "angle_thresh_DTM = 0.005\n",
    "angle_thresh_prosup = 0.001"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4d0aa407be124dd48587ee12a20add74",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=9.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(4392, 250, 8) (4392, 3)\n"
     ]
    }
   ],
   "source": [
    "X_prosup, y_prosup = multiple_prep_data_prosup(prosup_files_df,\n",
    "                                      window,\n",
    "                                      interval,\n",
    "                                      Normalize,\n",
    "                                      rms,\n",
    "                                      angle_thresh_prosup)\n",
    "print(X_prosup.shape,y_prosup.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3855216a15a04b1b9c75dc708719dfab",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=9.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "(4392, 250, 8) (4392, 3)\n"
     ]
    }
   ],
   "source": [
    "X_dtm, y_dtm = multiple_prep_data_DTM(dtm_files_df,\n",
    "                                      window,\n",
    "                                      interval,\n",
    "                                      Normalize,\n",
    "                                      rms,\n",
    "                                      angle_thresh_DTM)\n",
    "print(X_dtm.shape,y_dtm.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Merge ProSup and DTM Datasets to Bianry excluding Rest Labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "82c8814197d943f198f3bf08d52be48c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=4392.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "db00064a97b842eca6d330298850eb46",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=4392.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "X_dtm_n = []\n",
    "for i in tqdm(range(y_dtm.shape[0])):\n",
    "    if y_dtm[i][0]!=1:\n",
    "        X_dtm_n.append(X_dtm[i])\n",
    "X_dtm_n = np.array(X_dtm_n)        \n",
    "\n",
    "X_prosup_n = []\n",
    "for i in tqdm(range(y_prosup.shape[0])):\n",
    "    if y_prosup[i][0]!=1:\n",
    "        X_prosup_n.append(X_prosup[i])\n",
    "X_prosup_n = np.array(X_prosup_n)  \n",
    "\n",
    "# prosup - 0\n",
    "# DTM - 1\n",
    "X = np.vstack((X_prosup_n,X_dtm_n))\n",
    "a = np.array([[1,0]]*X_prosup_n.shape[0])\n",
    "b = np.array([[0,1]]*X_dtm_n.shape[0])\n",
    "y = np.vstack((a,b))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((6639, 250, 8), (6639, 2))"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('prepared_data/test_X',X)\n",
    "np.save('prepared_data/test_y',y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Time Window Normalization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [],
   "source": [
    "def variance(data, ddof=0):\n",
    "    n = len(data)\n",
    "    mean = sum(data) / n\n",
    "    return sum((x - mean) ** 2 for x in data) / (n - ddof)\n",
    "\n",
    "X_mean = np.zeros(X[0].shape)\n",
    "for x in X:\n",
    "    X_mean = X_mean + x\n",
    "X_mean = X_mean/X.shape[0]\n",
    "X_std = np.sqrt(variance(X))\n",
    "np.save('mean_std',np.array([X_mean,X_std]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_mean = np.load('mean_std_master.npy')[0]\n",
    "X_std = np.load('mean_std_master.npy')[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_mean = np.load('mean_std_master_norm.npy')[0]\n",
    "X_std = np.load('mean_std_master_norm.npy')[1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "530599d7b359485685f8aa5114d3b46c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=20317.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "Xn = []\n",
    "for x in tqdm(X):\n",
    "    Xn.append((x-X_mean)/X_std)\n",
    "X = np.array(Xn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('prepared_data/X_prosup_dtm_tw',X)\n",
    "np.save('prepared_data/y_prosup_dtm_tw',y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = np.load('prepared_data/X_prosup_dtm_tw.npy')\n",
    "y = np.load('prepared_data/y_prosup_dtm_tw.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(374616, 250, 8)\n",
      "(93654, 250, 8)\n",
      "(374616, 2)\n",
      "(93654, 2)\n"
     ]
    }
   ],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y,test_size=0.2)\n",
    "for data in [X_train, X_test, y_train, y_test]:\n",
    "    print(data.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save('prepared_data/X_train',X_train)\n",
    "np.save('prepared_data/X_test',X_test)\n",
    "np.save('prepared_data/y_train',y_train)\n",
    "np.save('prepared_data/y_test',y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(490274, 250, 8)\n",
      "(122569, 250, 8)\n",
      "(490274, 2)\n",
      "(122569, 2)\n"
     ]
    }
   ],
   "source": [
    "X_train = np.load('prepared_data/X_train.npy')\n",
    "X_test = np.load('prepared_data/X_test.npy')\n",
    "y_train = np.load('prepared_data/y_train.npy')\n",
    "y_test = np.load('prepared_data/y_test.npy')\n",
    "for data in [X_train, X_test, y_train, y_test]:\n",
    "    print(data.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_true = []\n",
    "for d in y_test:\n",
    "    idx = list(d).index(1)\n",
    "    y_true.append(idx)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "del X,y,dtm_files_df,prosup_files_df,X_dtm, y_dtm,X_prosup,y_prosup"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Label Distribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def count(y):\n",
    "    s0 = 0\n",
    "    s1 = 0\n",
    "    for arr in y:\n",
    "        if arr[0]==1:\n",
    "            s0 = s0+1\n",
    "        if arr[1]==1:\n",
    "            s1 = s1+1\n",
    "    ls = [s0,s1]\n",
    "    for i in range(2):\n",
    "        print(i,'-',ls[i],'[',ls[i]*100/len(y),'%]')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "count(y_train)\n",
    "count(y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 - 2271 [ 56.53472740851382 %]\n",
      "1 - 1746 [ 43.46527259148618 %]\n"
     ]
    }
   ],
   "source": [
    "count(y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# InceptionTime Trainer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import sklearn\n",
    "import pandas as pd\n",
    "import sys\n",
    "import subprocess\n",
    "import tensorflow as tf\n",
    "from InceptionTime.classifiers.inception import Classifier_INCEPTION"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 250, 8)]     0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1d (Conv1D)                 (None, 250, 32)      256         input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d (MaxPooling1D)    (None, 250, 8)       0           input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_1 (Conv1D)               (None, 250, 32)      10240       conv1d[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_2 (Conv1D)               (None, 250, 32)      5120        conv1d[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_3 (Conv1D)               (None, 250, 32)      2048        conv1d[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_4 (Conv1D)               (None, 250, 32)      256         max_pooling1d[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "concatenate (Concatenate)       (None, 250, 128)     0           conv1d_1[0][0]                   \n",
      "                                                                 conv1d_2[0][0]                   \n",
      "                                                                 conv1d_3[0][0]                   \n",
      "                                                                 conv1d_4[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization (BatchNorma (None, 250, 128)     512         concatenate[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation (Activation)         (None, 250, 128)     0           batch_normalization[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_5 (Conv1D)               (None, 250, 32)      4096        activation[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_1 (MaxPooling1D)  (None, 250, 128)     0           activation[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_6 (Conv1D)               (None, 250, 32)      10240       conv1d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_7 (Conv1D)               (None, 250, 32)      5120        conv1d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_8 (Conv1D)               (None, 250, 32)      2048        conv1d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_9 (Conv1D)               (None, 250, 32)      4096        max_pooling1d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 250, 128)     0           conv1d_6[0][0]                   \n",
      "                                                                 conv1d_7[0][0]                   \n",
      "                                                                 conv1d_8[0][0]                   \n",
      "                                                                 conv1d_9[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1 (BatchNor (None, 250, 128)     512         concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_1 (Activation)       (None, 250, 128)     0           batch_normalization_1[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_10 (Conv1D)              (None, 250, 32)      4096        activation_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_2 (MaxPooling1D)  (None, 250, 128)     0           activation_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_11 (Conv1D)              (None, 250, 32)      10240       conv1d_10[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_12 (Conv1D)              (None, 250, 32)      5120        conv1d_10[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_13 (Conv1D)              (None, 250, 32)      2048        conv1d_10[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_14 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_2 (Concatenate)     (None, 250, 128)     0           conv1d_11[0][0]                  \n",
      "                                                                 conv1d_12[0][0]                  \n",
      "                                                                 conv1d_13[0][0]                  \n",
      "                                                                 conv1d_14[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_15 (Conv1D)              (None, 250, 128)     1024        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_2 (BatchNor (None, 250, 128)     512         concatenate_2[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_3 (BatchNor (None, 250, 128)     512         conv1d_15[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "activation_2 (Activation)       (None, 250, 128)     0           batch_normalization_2[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "add (Add)                       (None, 250, 128)     0           batch_normalization_3[0][0]      \n",
      "                                                                 activation_2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_3 (Activation)       (None, 250, 128)     0           add[0][0]                        \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_16 (Conv1D)              (None, 250, 32)      4096        activation_3[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_3 (MaxPooling1D)  (None, 250, 128)     0           activation_3[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_17 (Conv1D)              (None, 250, 32)      10240       conv1d_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_18 (Conv1D)              (None, 250, 32)      5120        conv1d_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_19 (Conv1D)              (None, 250, 32)      2048        conv1d_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_20 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_3 (Concatenate)     (None, 250, 128)     0           conv1d_17[0][0]                  \n",
      "                                                                 conv1d_18[0][0]                  \n",
      "                                                                 conv1d_19[0][0]                  \n",
      "                                                                 conv1d_20[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_4 (BatchNor (None, 250, 128)     512         concatenate_3[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_4 (Activation)       (None, 250, 128)     0           batch_normalization_4[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_21 (Conv1D)              (None, 250, 32)      4096        activation_4[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_4 (MaxPooling1D)  (None, 250, 128)     0           activation_4[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_22 (Conv1D)              (None, 250, 32)      10240       conv1d_21[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_23 (Conv1D)              (None, 250, 32)      5120        conv1d_21[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_24 (Conv1D)              (None, 250, 32)      2048        conv1d_21[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_25 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_4[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_4 (Concatenate)     (None, 250, 128)     0           conv1d_22[0][0]                  \n",
      "                                                                 conv1d_23[0][0]                  \n",
      "                                                                 conv1d_24[0][0]                  \n",
      "                                                                 conv1d_25[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_5 (BatchNor (None, 250, 128)     512         concatenate_4[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_5 (Activation)       (None, 250, 128)     0           batch_normalization_5[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_26 (Conv1D)              (None, 250, 32)      4096        activation_5[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_5 (MaxPooling1D)  (None, 250, 128)     0           activation_5[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_27 (Conv1D)              (None, 250, 32)      10240       conv1d_26[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_28 (Conv1D)              (None, 250, 32)      5120        conv1d_26[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_29 (Conv1D)              (None, 250, 32)      2048        conv1d_26[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_30 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_5[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_5 (Concatenate)     (None, 250, 128)     0           conv1d_27[0][0]                  \n",
      "                                                                 conv1d_28[0][0]                  \n",
      "                                                                 conv1d_29[0][0]                  \n",
      "                                                                 conv1d_30[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_31 (Conv1D)              (None, 250, 128)     16384       activation_3[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_6 (BatchNor (None, 250, 128)     512         concatenate_5[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_7 (BatchNor (None, 250, 128)     512         conv1d_31[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "activation_6 (Activation)       (None, 250, 128)     0           batch_normalization_6[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "add_1 (Add)                     (None, 250, 128)     0           batch_normalization_7[0][0]      \n",
      "                                                                 activation_6[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_7 (Activation)       (None, 250, 128)     0           add_1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_32 (Conv1D)              (None, 250, 32)      4096        activation_7[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_6 (MaxPooling1D)  (None, 250, 128)     0           activation_7[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_33 (Conv1D)              (None, 250, 32)      10240       conv1d_32[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_34 (Conv1D)              (None, 250, 32)      5120        conv1d_32[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_35 (Conv1D)              (None, 250, 32)      2048        conv1d_32[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_36 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_6[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_6 (Concatenate)     (None, 250, 128)     0           conv1d_33[0][0]                  \n",
      "                                                                 conv1d_34[0][0]                  \n",
      "                                                                 conv1d_35[0][0]                  \n",
      "                                                                 conv1d_36[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_8 (BatchNor (None, 250, 128)     512         concatenate_6[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_8 (Activation)       (None, 250, 128)     0           batch_normalization_8[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_37 (Conv1D)              (None, 250, 32)      4096        activation_8[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_7 (MaxPooling1D)  (None, 250, 128)     0           activation_8[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_38 (Conv1D)              (None, 250, 32)      10240       conv1d_37[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_39 (Conv1D)              (None, 250, 32)      5120        conv1d_37[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_40 (Conv1D)              (None, 250, 32)      2048        conv1d_37[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_41 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_7[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_7 (Concatenate)     (None, 250, 128)     0           conv1d_38[0][0]                  \n",
      "                                                                 conv1d_39[0][0]                  \n",
      "                                                                 conv1d_40[0][0]                  \n",
      "                                                                 conv1d_41[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_9 (BatchNor (None, 250, 128)     512         concatenate_7[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_9 (Activation)       (None, 250, 128)     0           batch_normalization_9[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_42 (Conv1D)              (None, 250, 32)      4096        activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_8 (MaxPooling1D)  (None, 250, 128)     0           activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_43 (Conv1D)              (None, 250, 32)      10240       conv1d_42[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_44 (Conv1D)              (None, 250, 32)      5120        conv1d_42[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_45 (Conv1D)              (None, 250, 32)      2048        conv1d_42[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_46 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_8[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_8 (Concatenate)     (None, 250, 128)     0           conv1d_43[0][0]                  \n",
      "                                                                 conv1d_44[0][0]                  \n",
      "                                                                 conv1d_45[0][0]                  \n",
      "                                                                 conv1d_46[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_47 (Conv1D)              (None, 250, 128)     16384       activation_7[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_10 (BatchNo (None, 250, 128)     512         concatenate_8[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_11 (BatchNo (None, 250, 128)     512         conv1d_47[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "activation_10 (Activation)      (None, 250, 128)     0           batch_normalization_10[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_2 (Add)                     (None, 250, 128)     0           batch_normalization_11[0][0]     \n",
      "                                                                 activation_10[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_11 (Activation)      (None, 250, 128)     0           add_2[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_48 (Conv1D)              (None, 250, 32)      4096        activation_11[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_9 (MaxPooling1D)  (None, 250, 128)     0           activation_11[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_49 (Conv1D)              (None, 250, 32)      10240       conv1d_48[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_50 (Conv1D)              (None, 250, 32)      5120        conv1d_48[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_51 (Conv1D)              (None, 250, 32)      2048        conv1d_48[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_52 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_9[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_9 (Concatenate)     (None, 250, 128)     0           conv1d_49[0][0]                  \n",
      "                                                                 conv1d_50[0][0]                  \n",
      "                                                                 conv1d_51[0][0]                  \n",
      "                                                                 conv1d_52[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_12 (BatchNo (None, 250, 128)     512         concatenate_9[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_12 (Activation)      (None, 250, 128)     0           batch_normalization_12[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_53 (Conv1D)              (None, 250, 32)      4096        activation_12[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_10 (MaxPooling1D) (None, 250, 128)     0           activation_12[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_54 (Conv1D)              (None, 250, 32)      10240       conv1d_53[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_55 (Conv1D)              (None, 250, 32)      5120        conv1d_53[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_56 (Conv1D)              (None, 250, 32)      2048        conv1d_53[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_57 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_10[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_10 (Concatenate)    (None, 250, 128)     0           conv1d_54[0][0]                  \n",
      "                                                                 conv1d_55[0][0]                  \n",
      "                                                                 conv1d_56[0][0]                  \n",
      "                                                                 conv1d_57[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_13 (BatchNo (None, 250, 128)     512         concatenate_10[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_13 (Activation)      (None, 250, 128)     0           batch_normalization_13[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_58 (Conv1D)              (None, 250, 32)      4096        activation_13[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_11 (MaxPooling1D) (None, 250, 128)     0           activation_13[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_59 (Conv1D)              (None, 250, 32)      10240       conv1d_58[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_60 (Conv1D)              (None, 250, 32)      5120        conv1d_58[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_61 (Conv1D)              (None, 250, 32)      2048        conv1d_58[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_62 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_11[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_11 (Concatenate)    (None, 250, 128)     0           conv1d_59[0][0]                  \n",
      "                                                                 conv1d_60[0][0]                  \n",
      "                                                                 conv1d_61[0][0]                  \n",
      "                                                                 conv1d_62[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_63 (Conv1D)              (None, 250, 128)     16384       activation_11[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_14 (BatchNo (None, 250, 128)     512         concatenate_11[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_15 (BatchNo (None, 250, 128)     512         conv1d_63[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "activation_14 (Activation)      (None, 250, 128)     0           batch_normalization_14[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_3 (Add)                     (None, 250, 128)     0           batch_normalization_15[0][0]     \n",
      "                                                                 activation_14[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_15 (Activation)      (None, 250, 128)     0           add_3[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_64 (Conv1D)              (None, 250, 32)      4096        activation_15[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_12 (MaxPooling1D) (None, 250, 128)     0           activation_15[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_65 (Conv1D)              (None, 250, 32)      10240       conv1d_64[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_66 (Conv1D)              (None, 250, 32)      5120        conv1d_64[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_67 (Conv1D)              (None, 250, 32)      2048        conv1d_64[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_68 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_12[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_12 (Concatenate)    (None, 250, 128)     0           conv1d_65[0][0]                  \n",
      "                                                                 conv1d_66[0][0]                  \n",
      "                                                                 conv1d_67[0][0]                  \n",
      "                                                                 conv1d_68[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_16 (BatchNo (None, 250, 128)     512         concatenate_12[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_16 (Activation)      (None, 250, 128)     0           batch_normalization_16[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_69 (Conv1D)              (None, 250, 32)      4096        activation_16[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_13 (MaxPooling1D) (None, 250, 128)     0           activation_16[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_70 (Conv1D)              (None, 250, 32)      10240       conv1d_69[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_71 (Conv1D)              (None, 250, 32)      5120        conv1d_69[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_72 (Conv1D)              (None, 250, 32)      2048        conv1d_69[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_73 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_13[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_13 (Concatenate)    (None, 250, 128)     0           conv1d_70[0][0]                  \n",
      "                                                                 conv1d_71[0][0]                  \n",
      "                                                                 conv1d_72[0][0]                  \n",
      "                                                                 conv1d_73[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_17 (BatchNo (None, 250, 128)     512         concatenate_13[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_17 (Activation)      (None, 250, 128)     0           batch_normalization_17[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_74 (Conv1D)              (None, 250, 32)      4096        activation_17[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_14 (MaxPooling1D) (None, 250, 128)     0           activation_17[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_75 (Conv1D)              (None, 250, 32)      10240       conv1d_74[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_76 (Conv1D)              (None, 250, 32)      5120        conv1d_74[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_77 (Conv1D)              (None, 250, 32)      2048        conv1d_74[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_78 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_14[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_14 (Concatenate)    (None, 250, 128)     0           conv1d_75[0][0]                  \n",
      "                                                                 conv1d_76[0][0]                  \n",
      "                                                                 conv1d_77[0][0]                  \n",
      "                                                                 conv1d_78[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_79 (Conv1D)              (None, 250, 128)     16384       activation_15[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_18 (BatchNo (None, 250, 128)     512         concatenate_14[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_19 (BatchNo (None, 250, 128)     512         conv1d_79[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "activation_18 (Activation)      (None, 250, 128)     0           batch_normalization_18[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_4 (Add)                     (None, 250, 128)     0           batch_normalization_19[0][0]     \n",
      "                                                                 activation_18[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_19 (Activation)      (None, 250, 128)     0           add_4[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_80 (Conv1D)              (None, 250, 32)      4096        activation_19[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_15 (MaxPooling1D) (None, 250, 128)     0           activation_19[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_81 (Conv1D)              (None, 250, 32)      10240       conv1d_80[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_82 (Conv1D)              (None, 250, 32)      5120        conv1d_80[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_83 (Conv1D)              (None, 250, 32)      2048        conv1d_80[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_84 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_15[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_15 (Concatenate)    (None, 250, 128)     0           conv1d_81[0][0]                  \n",
      "                                                                 conv1d_82[0][0]                  \n",
      "                                                                 conv1d_83[0][0]                  \n",
      "                                                                 conv1d_84[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_20 (BatchNo (None, 250, 128)     512         concatenate_15[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_20 (Activation)      (None, 250, 128)     0           batch_normalization_20[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_85 (Conv1D)              (None, 250, 32)      4096        activation_20[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_16 (MaxPooling1D) (None, 250, 128)     0           activation_20[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_86 (Conv1D)              (None, 250, 32)      10240       conv1d_85[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_87 (Conv1D)              (None, 250, 32)      5120        conv1d_85[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_88 (Conv1D)              (None, 250, 32)      2048        conv1d_85[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_89 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_16[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_16 (Concatenate)    (None, 250, 128)     0           conv1d_86[0][0]                  \n",
      "                                                                 conv1d_87[0][0]                  \n",
      "                                                                 conv1d_88[0][0]                  \n",
      "                                                                 conv1d_89[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_21 (BatchNo (None, 250, 128)     512         concatenate_16[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_21 (Activation)      (None, 250, 128)     0           batch_normalization_21[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_90 (Conv1D)              (None, 250, 32)      4096        activation_21[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_17 (MaxPooling1D) (None, 250, 128)     0           activation_21[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_91 (Conv1D)              (None, 250, 32)      10240       conv1d_90[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_92 (Conv1D)              (None, 250, 32)      5120        conv1d_90[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_93 (Conv1D)              (None, 250, 32)      2048        conv1d_90[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_94 (Conv1D)              (None, 250, 32)      4096        max_pooling1d_17[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_17 (Concatenate)    (None, 250, 128)     0           conv1d_91[0][0]                  \n",
      "                                                                 conv1d_92[0][0]                  \n",
      "                                                                 conv1d_93[0][0]                  \n",
      "                                                                 conv1d_94[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_95 (Conv1D)              (None, 250, 128)     16384       activation_19[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_22 (BatchNo (None, 250, 128)     512         concatenate_17[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_23 (BatchNo (None, 250, 128)     512         conv1d_95[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "activation_22 (Activation)      (None, 250, 128)     0           batch_normalization_22[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_5 (Add)                     (None, 250, 128)     0           batch_normalization_23[0][0]     \n",
      "                                                                 activation_22[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_23 (Activation)      (None, 250, 128)     0           add_5[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_96 (Conv1D)              (None, 250, 32)      4096        activation_23[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_18 (MaxPooling1D) (None, 250, 128)     0           activation_23[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_97 (Conv1D)              (None, 250, 32)      10240       conv1d_96[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_98 (Conv1D)              (None, 250, 32)      5120        conv1d_96[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_99 (Conv1D)              (None, 250, 32)      2048        conv1d_96[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_100 (Conv1D)             (None, 250, 32)      4096        max_pooling1d_18[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_18 (Concatenate)    (None, 250, 128)     0           conv1d_97[0][0]                  \n",
      "                                                                 conv1d_98[0][0]                  \n",
      "                                                                 conv1d_99[0][0]                  \n",
      "                                                                 conv1d_100[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_24 (BatchNo (None, 250, 128)     512         concatenate_18[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_24 (Activation)      (None, 250, 128)     0           batch_normalization_24[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_101 (Conv1D)             (None, 250, 32)      4096        activation_24[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_19 (MaxPooling1D) (None, 250, 128)     0           activation_24[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_102 (Conv1D)             (None, 250, 32)      10240       conv1d_101[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_103 (Conv1D)             (None, 250, 32)      5120        conv1d_101[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_104 (Conv1D)             (None, 250, 32)      2048        conv1d_101[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_105 (Conv1D)             (None, 250, 32)      4096        max_pooling1d_19[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_19 (Concatenate)    (None, 250, 128)     0           conv1d_102[0][0]                 \n",
      "                                                                 conv1d_103[0][0]                 \n",
      "                                                                 conv1d_104[0][0]                 \n",
      "                                                                 conv1d_105[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_25 (BatchNo (None, 250, 128)     512         concatenate_19[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_25 (Activation)      (None, 250, 128)     0           batch_normalization_25[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling1d (Globa (None, 128)          0           activation_25[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 2)            258         global_average_pooling1d[0][0]   \n",
      "==================================================================================================\n",
      "Total params: 600,834\n",
      "Trainable params: 594,178\n",
      "Non-trainable params: 6,656\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "clf = Classifier_INCEPTION('', (X_train.shape[1],X_train.shape[2]),nb_classes=2,\n",
    "                            verbose=True,batch_size=128,nb_epochs=25,nb_filters=32,\n",
    "                            depth=20, kernel_size=11)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n"
     ]
    },
    {
     "ename": "ResourceExhaustedError",
     "evalue": " OOM when allocating tensor with shape[128,32,1,250] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc\n\t [[node model/conv1d_10/conv1d (defined at F:\\The Stuffs\\Awear\\Final_Project\\wrist_control_CNN_AWEAR\\InceptionTime\\classifiers\\inception.py:122) ]]\nHint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info. This isn't available when running in Eager mode.\n [Op:__inference_train_function_17157]\n\nFunction call stack:\ntrain_function\n",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mResourceExhaustedError\u001b[0m                    Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-9-18f29238e70f>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mdf_metrics\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mclf\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_train\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX_test\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_test\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my_true\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mplot_test_acc\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32mF:\\The Stuffs\\Awear\\Final_Project\\wrist_control_CNN_AWEAR\\InceptionTime\\classifiers\\inception.py\u001b[0m in \u001b[0;36mfit\u001b[1;34m(self, x_train, y_train, x_val, y_val, y_true, plot_test_acc)\u001b[0m\n\u001b[0;32m    120\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mplot_test_acc\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    121\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 122\u001b[1;33m             hist = self.model.fit(x_train, y_train, batch_size=mini_batch_size, epochs=self.nb_epochs,\n\u001b[0m\u001b[0;32m    123\u001b[0m                                   verbose=self.verbose, validation_data=(x_val, y_val), callbacks=self.callbacks, shuffle=True)\n\u001b[0;32m    124\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\keras\\engine\\training.py\u001b[0m in \u001b[0;36mfit\u001b[1;34m(self, x, y, batch_size, epochs, verbose, callbacks, validation_split, validation_data, shuffle, class_weight, sample_weight, initial_epoch, steps_per_epoch, validation_steps, validation_batch_size, validation_freq, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[0;32m   1182\u001b[0m                 _r=1):\n\u001b[0;32m   1183\u001b[0m               \u001b[0mcallbacks\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mon_train_batch_begin\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mstep\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1184\u001b[1;33m               \u001b[0mtmp_logs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtrain_function\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0miterator\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1185\u001b[0m               \u001b[1;32mif\u001b[0m \u001b[0mdata_handler\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshould_sync\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1186\u001b[0m                 \u001b[0mcontext\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0masync_wait\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\eager\\def_function.py\u001b[0m in \u001b[0;36m__call__\u001b[1;34m(self, *args, **kwds)\u001b[0m\n\u001b[0;32m    883\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    884\u001b[0m       \u001b[1;32mwith\u001b[0m \u001b[0mOptionalXlaContext\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_jit_compile\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 885\u001b[1;33m         \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_call\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    886\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    887\u001b[0m       \u001b[0mnew_tracing_count\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mexperimental_get_tracing_count\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\eager\\def_function.py\u001b[0m in \u001b[0;36m_call\u001b[1;34m(self, *args, **kwds)\u001b[0m\n\u001b[0;32m    948\u001b[0m         \u001b[1;31m# Lifting succeeded, so variables are initialized and we can run the\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    949\u001b[0m         \u001b[1;31m# stateless function.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 950\u001b[1;33m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_stateless_fn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwds\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    951\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    952\u001b[0m       \u001b[0m_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfiltered_flat_args\u001b[0m \u001b[1;33m=\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\eager\\function.py\u001b[0m in \u001b[0;36m__call__\u001b[1;34m(self, *args, **kwargs)\u001b[0m\n\u001b[0;32m   3037\u001b[0m       (graph_function,\n\u001b[0;32m   3038\u001b[0m        filtered_flat_args) = self._maybe_define_function(args, kwargs)\n\u001b[1;32m-> 3039\u001b[1;33m     return graph_function._call_flat(\n\u001b[0m\u001b[0;32m   3040\u001b[0m         filtered_flat_args, captured_inputs=graph_function.captured_inputs)  # pylint: disable=protected-access\n\u001b[0;32m   3041\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\eager\\function.py\u001b[0m in \u001b[0;36m_call_flat\u001b[1;34m(self, args, captured_inputs, cancellation_manager)\u001b[0m\n\u001b[0;32m   1961\u001b[0m         and executing_eagerly):\n\u001b[0;32m   1962\u001b[0m       \u001b[1;31m# No tape is watching; skip to running the function.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1963\u001b[1;33m       return self._build_call_outputs(self._inference_function.call(\n\u001b[0m\u001b[0;32m   1964\u001b[0m           ctx, args, cancellation_manager=cancellation_manager))\n\u001b[0;32m   1965\u001b[0m     forward_backward = self._select_forward_and_backward_functions(\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\eager\\function.py\u001b[0m in \u001b[0;36mcall\u001b[1;34m(self, ctx, args, cancellation_manager)\u001b[0m\n\u001b[0;32m    589\u001b[0m       \u001b[1;32mwith\u001b[0m \u001b[0m_InterpolateFunctionError\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    590\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mcancellation_manager\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 591\u001b[1;33m           outputs = execute.execute(\n\u001b[0m\u001b[0;32m    592\u001b[0m               \u001b[0mstr\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msignature\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mname\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    593\u001b[0m               \u001b[0mnum_outputs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_num_outputs\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\eager\\execute.py\u001b[0m in \u001b[0;36mquick_execute\u001b[1;34m(op_name, num_outputs, inputs, attrs, ctx, name)\u001b[0m\n\u001b[0;32m     57\u001b[0m   \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     58\u001b[0m     \u001b[0mctx\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mensure_initialized\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 59\u001b[1;33m     tensors = pywrap_tfe.TFE_Py_Execute(ctx._handle, device_name, op_name,\n\u001b[0m\u001b[0;32m     60\u001b[0m                                         inputs, attrs, num_outputs)\n\u001b[0;32m     61\u001b[0m   \u001b[1;32mexcept\u001b[0m \u001b[0mcore\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_NotOkStatusException\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mResourceExhaustedError\u001b[0m:  OOM when allocating tensor with shape[128,32,1,250] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc\n\t [[node model/conv1d_10/conv1d (defined at F:\\The Stuffs\\Awear\\Final_Project\\wrist_control_CNN_AWEAR\\InceptionTime\\classifiers\\inception.py:122) ]]\nHint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info. This isn't available when running in Eager mode.\n [Op:__inference_train_function_17157]\n\nFunction call stack:\ntrain_function\n"
     ]
    }
   ],
   "source": [
    "df_metrics = clf.fit(X_train, y_train, X_test, y_test, y_true,plot_test_acc=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\moham\\AppData\\Local\\Programs\\Python\\Python38\\lib\\site-packages\\tensorflow\\python\\keras\\utils\\generic_utils.py:494: CustomMaskWarning: Custom mask layers require a config and must override get_config. When loading, the custom mask layer must be passed to the custom_objects argument.\n",
      "  warnings.warn('Custom mask layers require a config and must override '\n"
     ]
    }
   ],
   "source": [
    "best_model = load_model('best_model.hdf5')\n",
    "best_model.save('models/prosup_dtm_{}_norm.hdf5'.format(window))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 - 1104 [ 59.037433155080215 %]\n",
      "1 - 766 [ 40.962566844919785 %]\n"
     ]
    }
   ],
   "source": [
    "count(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "54/54 [==============================] - 3s 28ms/step - loss: 0.5043 - accuracy: 0.9194\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.504290759563446, 0.9194395542144775]"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_model = load_model('best_models/prosup_dtm_250/prosup_dtm_250_norm_99.hdf5')\n",
    "best_model.evaluate(X,y) # Subject - Varun"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "yt = []\n",
    "for e in y:\n",
    "    if e[0] == 1:\n",
    "        yt.append('ProSup')\n",
    "    else:\n",
    "        yt.append('DTM')\n",
    "yp = []\n",
    "pred = best_model.predict(X)\n",
    "for e in pred:\n",
    "    if e[0] == 1:\n",
    "        yp.append('ProSup')\n",
    "    else:\n",
    "        yp.append('DTM')\n",
    "\n",
    "from sklearn.metrics import confusion_matrix\n",
    "matrix = confusion_matrix(yt, yp,labels=['ProSup','DTM'])\n",
    "\n",
    "group_names = ['True Neg','False Pos','False Neg','True Pos']\n",
    "\n",
    "group_counts = ['{0:0.0f}'.format(value) for value in\n",
    "                matrix.flatten()]\n",
    "group_percentages = ['{0:.2%}'.format(value) for value in\n",
    "                     matrix.flatten()/np.sum(matrix)]\n",
    "labels = [f'{v1}\\n{v2}\\n{v3}' for v1, v2, v3 in\n",
    "          zip(group_names,group_counts,group_percentages)]\n",
    "labels = np.asarray(labels).reshape(2,2)\n",
    "sns.heatmap(matrix, annot=labels, fmt='', cmap='Blues')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "metrics_df = pd.read_csv('df_metrics.csv')\n",
    "best_model_df = pd.read_csv('df_best_model.csv')\n",
    "history_df = pd.read_csv('history.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>precision</th>\n",
       "      <th>accuracy</th>\n",
       "      <th>recall</th>\n",
       "      <th>duration</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.997051</td>\n",
       "      <td>0.997143</td>\n",
       "      <td>0.996341</td>\n",
       "      <td>13798.457582</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   precision  accuracy    recall      duration\n",
       "0   0.997051  0.997143  0.996341  13798.457582"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metrics_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Loss'}>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "history_df[['accuracy','val_accuracy']].plot(figsize=(14,8),title='Accuracy',grid=True,fontsize=15)\n",
    "history_df[['loss','val_loss']].plot(figsize=(14,8),title='Loss',grid=True,fontsize=15)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.8.0 64-bit",
   "language": "python",
   "name": "python38064bitcaed634bb72a48a5bf9eda0dc4f94389"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}