exploratory-analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### README\n",
    "The purpose of this notebook is to walk the reader through some of the exploratory analysis that was done when working on this project. This is by no means all the work that was done, but showcases the things that we thought were important or significant. We have tried to explain our thought process in an organic and natural way"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import required libraries\n",
    "import pandas as pd # for holding data in a dataframe\n",
    "import os # for manipulating directories and listing files\n",
    "from matplotlib import pyplot as plt #visualization\n",
    "import numpy as np # linear algebra and array work\n",
    "from statsmodels.stats.proportion import proportions_ztest #significance testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_csvs(dirPath):\n",
    "    '''\n",
    "    This function reads all the .csv files at the specified director \n",
    "    and then loads them individually into a merged dataframe. It \n",
    "    returns a merged dataframe with a reset index. It also appends the video number\n",
    "    into the dataframe.\n",
    "    '''\n",
    "    # list files in directory\n",
    "    files = os.listdir(dirPath)\n",
    "    \n",
    "    # Create list of valid .csv files in directory and sort numerically\n",
    "    csvs = [csv for csv in files if csv.endswith('.csv')]\n",
    "    csvs.sort(key = lambda x: int(x[1:-4]))\n",
    "    \n",
    "    # Parse through csv files\n",
    "    for inx,csv in enumerate(csvs):\n",
    "        # If this is the first file, we need to create the merged_df\n",
    "        if inx == 0:\n",
    "            merged_df = pd.read_csv(dirPath + csv)\n",
    "            merged_df['video'] = 0 \n",
    "        # else we just read and append\n",
    "        else :\n",
    "            temp_df = pd.read_csv(dirPath + csv)\n",
    "            temp_df['video'] = inx\n",
    "            merged_df = merged_df.append(temp_df)\n",
    "\n",
    "    return merged_df.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load happy data\n",
    "happyPath = 'dataset/happy_frames_openface/'\n",
    "happy_df = read_csvs(happyPath)\n",
    "\n",
    "# Load nervous data\n",
    "nervousPath = 'dataset/nervous_frames_openface/'\n",
    "nervous_df = read_csvs(nervousPath)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create list of AUs that we want to extract from the imported dataset\n",
    "AUs = [' AU01_r', ' AU02_r', ' AU04_r', ' AU05_r', ' AU06_r', ' AU07_r',\n",
    "       ' AU09_r', ' AU10_r', ' AU12_r', ' AU14_r', ' AU15_r', ' AU17_r',\n",
    "       ' AU20_r', ' AU23_r', ' AU25_r', ' AU26_r', ' AU45_r', 'video']\n",
    "# Create a pointer to the original df that only includes the columns we want\n",
    "happy_au_df = happy_df.loc[:,AUs]\n",
    "nervous_au_df = nervous_df.loc[:, AUs]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get the max AU intenisty per AU per video\n",
    "happy_max_au = happy_au_df.groupby('video').max()\n",
    "nervous_max_au = nervous_au_df.groupby('video').max() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Count the number of times the max AU intensity was larger than 2.5\n",
    "happy_au_arr = (happy_max_au > 2.5).sum()\n",
    "nervous_au_arr = (nervous_max_au > 2.5).sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAoAAAAHbCAYAAABbbCuKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALiAAAC4gB5Y4pSQAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZhcZZn38e+dFWSNMKwJBoQBY5AAMoICAXXglUUcFkdFCIIoisPI4CiD6MioiBvgIL4qhsWADr6AigYdFwgQHBCEhCWgIAQSBQZiFhYh2/3+Uadj0XbS1Umfqk4/38911ZU66/2c6lPdvzxni8xEkiRJ5RjS6QZIkiSpvQyAkiRJhTEASpIkFcYAKEmSVBgDoCRJUmEMgJIkSYUxAEqDWET8JCI+2Ol2rMxAb5/6T0TsExFzm4anRcSHO9kmqWQGQGkAioiLIyIj4lV9WObSiDi/eVxmviUzv9b/LVz5H/Cq3RNaWUdz+yJiv4hY0ELdDSLivIiYExF/jojfR8QnI2JY37di8Kg+v4yIZ6vXUxHxnYh4+Rqs87SI+F1EPFOt7xcRMXZ11pWZN2fm6NVtS1Obvt60jS9ExLKm4WcjYps1rSGVwAAoDTARsT7wduBPwAkdbs6AEhHDgf8GdgHeDKwPHAUcDny3A+0ZaKFzYWaun5nrA38LbAp8fnVWFBHvBv4JODwzNwB2AL4JdPTpAZl5UtM2ngTc0zVcvR7rZPuktYUBUBp43gE8B3wMOLYKPQBExJCIOCUiHqh6ZR6MiP8TEacARwMfrHpB7qvmf0kvXUQcEBF3RcTCiLgzIt7cNO3SiLgoIv6rWvdvI2K/NdmQ3tbZ1b6I2AT4CbBRU0/OPj2s8mgaweZtmfnbzFyWmXcC/wAc1m3d74yImRGxKCIejYjjepvWvRc1IjauetXGNk2fHBHfi4hFwAciYnhE/EfVEzkvIq6NiK2a1pERcVJE3FvVuzYiNmqavkM17qmI+FNEXNM07ZUR8aNq2qMRcWZEtPR7OzPnAz8AXl2t67CIeDgiomn9e0XE/IhYp4dV7An8MjPvrda3IDO/l5mPVsseFxEzIuKsiHg6Ip6IiH+MiDdU27qw+qyGVPOvsoc3InaLiBuqz+ChiDixle1sWn6V29fU3rOrn9Nj0e30g4h4R0TcHRELIuL2iHh9X9ogrU0MgNLAcwJwBfBfwMuAQ5umfQj4MI0gtCHwJuDRzPzPapmvVb0gr+6+0oh4JfBD4NPAJsDZwLURsW3TbO+g0cuzMTAFuLQftqfXdWbmPOAtNPVgZebNPazrQOC6zFzUbflHgNuAAwAi4lDgq8CpVd09gJm9TWvRO4HJ1bKTgc8CbwD2BrYEfkfjZ9fsH2n8rLYBRle1iYj1gF8A9wJjgS2AC6pp6wK/BK4Htgb2ofFZvqeVRkbEpjR6Rrs+x6nAusDEptmOA76TmS/0sIrpwNsj4uNVqOspJL4aWFC1+xM0fs7/UtUYBxwCvK2Ftm4B/Bz4v8DfVMucFRFv6m3ZJq1s33gaPZhb0viZnBMR+1ZtOAj4UrXMy4HPAT+q/nMiDToGQGkAiYhxNHpeLsvMZ4Hv89LDwB8APpWZv8mGxzLz/hZX/w5gWmZek5lLM/MqGn/k39k0z9TMvD4zlwGXAK/ohz+A/bnOTYE/rmTaH2mEB4APAl+p6i7PzP/NzLtamNaKn2Xmf2fmcuDP1fr+JTMfz8zFwJnAGyJiTNMyn8/MJzNzAXA1sHs1/hBgCfDxzHwuMxdn5g1N0+Zn5nnV+MeArwDvWkXbNqp6rxYA/0sjOF4AkJlLgW/TCDhUge7tNH4mfyUz/4tG2Hw9jXA1r+rNXa9ptqer9i2l8R+QDYGLMnNeZv4BuBHYbRXt7XIMcFPVw7is6nW8pJdt7d7eVrbvORrfn8WZ+T9Vm4+tpp0MfDEz76z2i2uAB4CDWm2DtDYxAEoDywnAzMzs6pG6DDgwIrauhl8BPLia6x4NzO427uFqfJcnmt4/V/27wUrWtwQY3jwi/nK4eslqrrM3TwNbrWTaVsBT1ftVfU5r8hkCNJ9jtimwHnBTU/B6AlgMNAfA7p9B1/a/Avh9ZvZ0Xt1YYHzXeqt1f5lGb9vKLMzMjTNzYxq9YZOrtnX13l0MHBGN80z/AZibmXesbGWZeVVmHgyMotH7egDw8aZZnmya9/ketvV5Gudp9mYscFC3bT2FRk9dX/S2fX/MzOZ981EaIbmrDWd3a8OEpunSoGIAlAaIKjwdA/xtdT7VEzR6KIZS9WrQ+IO1/UpWsbyXEnNp/JFrtm01fnU8Wi3f7JVN0/qqt/ZD4zDhQRGxYfPI6jD231XTu+qv7HNa1bRnaRx279JTAGlu5zwaIed1XcGreq2bmb9a9aasaMsrm89bazIH+E239W7Y0+H9nmTmi8DXafyMXl2N+y2Nw91H0tinLm5xXZmZ04GrgJ1bWaaP5gDf77atG2Rmn3rfWti+rZr+kwKNQ/J/aGrDad3asF5mnrNaWyQNcAZAaeB4K41DaLvR6HmYQONq108Dx1ch4RvAv0fEhGjYJv5yq5gnge1Wsf4rgf2qk+WHRsThNM4r636+WqsuB95Vndw/NCI2A74A/LA6fN1XTwIbRMTfrGKey4HfAz+IiB2rurvROFQ+tenw6TeAf46IidG4cGaziNi1hWl30uhx3TIiNgD+fVUNrg4Dfx34ctch34jYJCL+scVtngqMBP4jItaLiBERsX817cfA5hHxweoihqHVNu/XyoqjcYXyiTQC6sNNkyYDpwH70vg8V7b8e6p9ZeNqeDxwGNBKsO2rKcAbI+KIaFxUM7zax/dYjXWtavvWAz5Rfc6vo3Eu7RXVtK8C/xoRu1ffrZdFxJsjYo1vXSMNRAZAaeA4AfhuZj6QmU90vYD/pHF4c//q/f8Fvgc8Q+MCgq77nn0L2Lq66vHu7ivPzIdoXBRwFjAf+CTwD5n5cPd5W5GZ02jchuN8Gres+Q2NEHf8aq7vtzT+eN9fHYLbu4d5FgN/D9xD4+KI52j0Sv2Qxkn9XfP9gMbFCBcCC4HbqXquVjWNRmC4kca5XzNoBLTe/BvwP8D1EfEMjc/hgBa3+Vkat7PZncah5cdpnIvWPO1NNA7dzwO+w6oPAa+4iprG4fKjgEOrK4K7fI/GoeefZuZTPa2ksoBGkHq42q4f0LjVzhda2ba+qM4XPBB4P43P4EkaP58NV7XcSqxq++4FhlU1rqJx7uUNVRt+DJwOXETj+/EI8M/4d1KDVPR86okkabCKiN8Dp2bmtZ1uSx162r5o3Ornw5nZ0k3KpcHO/9lIUkEi4h00esFa6d1c6wz27ZP6y0C7i70kqSYRcT+Ne9xNqm7LM6gM9u2T+pOHgCVJkgrjIWBJkqTCGAAlSZIKM+jOAdxwww1z9Ghv2yRJksp2//33P5OZPd5OadAFwNGjRzNr1qxON0OSJKmjImKlT3ryELAkSVJhDICSJEmFGXSHgCVJ0trL29P1XeNR8X1jAJQkSR23ZMkS5syZw4svvtjppqx1Ro4cyZgxYxg+fHjLyxgAJUlSx82ZM4cNNtiAsWPHrlaPVqkyk3nz5jFnzhy22267lpczAEqSpI7KTF588UXGjh3LkCFentAXEcEmm2zC008/TWa2HJ79lCVJ0oBgz9/qWZ3PzQAoSZJUGA8BS5KkAWfs6VP7fZ2zzzm413mGDRvG0qVLVwxPnz6dM888k2nTpvV7ezrJHkBJkqTCGAAlSZJaMGfOHCZOnMhuu+3GzjvvzOWXX75i2tixYznjjDPYddddGTduHDfccAMA06ZN4/Wvfz1HHnkk48aN45BDDmHBggU8+uij7Ljjjivue/jCCy+w1VZbMX/+/LZsi4eAJUmSKsuWLWPChAkrhp977jm23nprADbddFN++tOfsu6667Jo0SJ23313DjnkEDbeeGMAli9fzl133cXMmTM59NBDeeihhwC4/fbbueuuuxg/fjwf+chH+MxnPsOXvvQltt9+e2644Qbe+MY3cvXVV/PGN76RUaNGtWU77QGUJEmqDB06lBkzZqx4XXLJJSumLV26lJNPPpmdd96ZffbZh8cff3xFyAM49thjAdhll13YZJNNePDBBwHYddddGT9+PADHHHMMN954IwDvf//7ueiiiwC46KKLOPHEE9uyjWAAlCRJasm5557Luuuuy4wZM5g5cyY77rgjL7zwwkrnX9ntWbrGH3zwwdx+++3ceuutPPHEE0ycOLGWdvfEAChJktSCRYsWscUWWzB06FBuu+02Zs6c+ZLp3/72twGYOXMm8+bNY/vttwfgrrvu4t577wVgypQp7LvvvkCjt/Hd7343Rx11FCeccEIbt8RzACVJ0gDUyi1b2u1DH/oQRxxxBNdeey3jxo1jjz32eMn0l73sZey1114sXLiQSy+9lBEjRgCw55578vnPf567776bMWPGvOTikUmTJnH22WczadKktm6LAVCSJKnSfA9AgL333nvFPQC33XZb7rzzzpUue/zxx/PJT37yr8aPHDmSKVOm9LjMrbfeyhFHHMFmm222+o1eDQZAaS3Uyg1SB+L/niVJf3H00Udz++23M3Vq/9/0ujcGQEmSpDU0e/bsHsfvt99+7Lfffj1Ou+KKK+prUC+8CESSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjBeBSJKkgedTG9WwzoW9zhIRTJo0iUsvvRSA6dOnc+aZZ664FcxgYQ+gJElSZciQIUybNm3Fc3xXx7Jly/qxRfUwAEqSJFUigo9+9KOcddZZPU6/4IIL+Lu/+zt22WUX3vve97JkyRIAxo4dy6c//WkmTpzI2WefzVZbbfWS5wS/5jWv4Z577mHx4sV84AMfYPz48YwfP55vfetbQOM2Ml2PjgO4/PLLOe644wC45ZZb2H333ZkwYQLjx4/ntttuW+PtNABKkiQ1ee9738stt9zCAw888JLx119/Pbfddhu33norM2fOZMSIESsCHMDixYu58cYb+cQnPsHEiRNX3OC56znAO++8M1//+td56qmnuPvuu7nppps4++yzue+++1bZni984Qucf/75zJgxgxkzZvDqV796jbfRAChJktRkxIgRnH766X/VC3jddddx8803s9tuuzFhwgSuv/56Hn744RXTjz766Je877rR8xVXXLFi2rRp03jPe97DkCFDePnLX85b3/pWbrzxxlW2Z9999+W0007j3HPP5fe//z3rr7/+Gm+jAVCSJKmb448/nl//+tcv6Z3LTE499dQVPXEPPPAAX/ziF1dMbw5mBx54IL/+9a+ZP38+V155Je9617tWrKO7iGDYsGEsX758xbgXX3xxxfvTTjuNKVOmMGzYMA477DCuvPLKNd4+A6AkSVI3w4cP54wzzuDss89eMe4tb3kLl1xyCfPnzwdgwYIFPPLIIytd/rDDDuPUU09lm222YcyYMQDsv//+XHLJJSxfvpw//elPXHvttey7775svvnmzJ8/nyeffJJly5bxwx/+cMW6HnzwQXbccUdOOeUUjj32WO6444413j5vAyNJktSDSZMm8bnPfW7F8Jvf/GZOOukkJk6cCMCwYcP4yle+wrbbbtvj8kcffTRveMMb+OY3v7li3EknncT999/Pa17zGgDOOOOMFef0ffazn2WvvfZiu+22Y6edduLZZ58F4Pzzz2fatGmMGDGCUaNGcdlll63xtkVPXZFrs3HjxuWsWbM63QypVmNPn9rrPLPPObgNLZGkNZeZPPDAA+y0005ERKebs9ZZ2ecXEfdn5rielvEQsCRJUmEMgJIkSYUxAEqSJBXGAChJkgaEwXZdQruszufmVcCSJKmjIoKRI0cyb948NtlkEy8E6YPMZN68eYwcObJPn5sBUJIkddyYMWOYM2cOTz/9dKebstYZOXLkivsMtsoAKEmSOm748OFst912HgZeDavTY2oAlCRJA4aHf9vDi0AkSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKU2sAjIj9I2JWRDwUERdHxLBu08dExC8j4v6IuDciPtNt+heqZX8XEUfW2VZJkqRS1BYAI2IoMBk4KjO3B9YHjuk221LgY5n5KmA3YJ+IeGu1/AHAnsBOwP7AeRGxQV3tlSRJKkWdPYB7AHMz875qeDJwePMMmfl4Zt5RvV8M3A28opp8OHBpZi7NzD8A04EDamyvJElSEeoMgKOBOU3Dj1XjehQRmwBvA37Rl+Uj4uTqMPOsiJg1f/78NW64JEnSYFb3RSDZ9D5WNlNEjASuAs7NzPv7snxmXpiZ47peo0aNWqMGS5IkDXZ1BsA5wDZNw6OBud1nqs4VvAK4IzPP6+vykiRJ6ps6A+AdwNYRMa4aPh64pof5vgksAj7abfw1wKSIGBoRWwF7Az+rq7GSJEmlqC0AZuYy4ETgqoh4CHgemBIRr42I6wAi4g00guEewF0RMSMiTqmW/zlwG/Bb4EbgtMx8pq72SpIklWJY77Osvsy8HhjXbfQdwEHV9FtYxbmBmflR/rpnUJIkSWvAJ4FIkiQVxgAoSZJUGAOgJElSYQyAkiRJhTEASpIkFcYAKEmSVBgDoCRJUmEMgJIkSYUxAEqSJBWm1ieBSNJgM/b0qS3NN/ucg2tuiSStPnsAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgozrNMNkFSTT23U4nwL622HJGnAsQdQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKU2sAjIj9I2JWRDwUERdHxLAe5rkgIuZGxNJu4/eLiGciYkb1uqrOtkqSJJWitgAYEUOBycBRmbk9sD5wTA+zXgnsvpLV3JaZE6rXkTU1VZIkqSh19gDuAczNzPuq4cnA4d1nyszpmflkje2QJElSkzoD4GhgTtPwY9W4vti9Ovx7U0Qc2H9NkyRJKtdfnZPXz7LpffRx2TuBV2TmooiYAFwXEXtl5qPNM0XEycDJXcNbbLHFajdWkiSpBHX2AM4BtmkaHg3MbXXhzFyUmYuq9zOAXwETepjvwswc1/UaNWrUGjZbkiRpcKszAN4BbB0R46rh44FrWl04IraMiKjejwZeB8zq91ZKkiQVprYAmJnLgBOBqyLiIeB5YEpEvDYiruuaLyK+ERFzgaHV7WC+UU06Arg3ImYAPwY+lpkP1tVeSZKkUtR6DmBmXg+M6zb6DuCgpnnev5Jlvwp8tb7WSZIklckngUiSJBXGAChJklSYum8DI0mS1D6f2qiFeRbW344Bzh5ASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqzLBWZ4yIrYCxzctk5k01tEmSJEk1aikARsRngeOB+4Fl1egEDqipXZIkSapJqz2AbwdemZnP19kYSZIk1a/VcwBnA0trbIckSZLapNUewLnAryJiKvBi18jMPLuWVkmSJKk2rQbA2dULYHgtLZEkSVJbtBQAM/OsuhsiSZKk9mjpHMCI2CwivhsRT0TE4xHxnYjYrO7GSZIkqf+1ehHIZOB2YEdgJ+A3wMV1NUqSJEn1aTUAjsnMczNzYfX6MjC6zoZJkiSpHq0GwCURsUvXQERMAJbU0yRJkiTVqdWrgE8FfhQRjwMBbA68q7ZWSZIkqTatXgU8PSK2p3EOYAAPZObiWlsmSZKkWqwyAEbEppn5dERsVY2aV/27aUSQmX+st3mSJEnqb731AF4KHALcAiSN3r8uCWxXT7MkSZJUl1UGwMw8pPp32/Y0R5IkSXVr9UbQP+th3HX93xxJkiTVrdXbwGzaPBAR6wJj+r85kiRJqtsqA2BEfCIilgATImJx9VoCPA5c05YWSpIkqV/1dg7gp4FPR8S5mfkvbWqTNOCNPX1qS/PNPufgmlsiSVLf9XoIOCKGAAe0oS2SJElqg14DYGYuBx6IiL9pQ3skSZJUs1YfBRfAvRHxS+DZrpGZ+b5aWiVJkqTatBoAr61ekiRJWsu1+izgywAiYp1q+IU6GyVJkqT6tHoj6B0i4n+AucAfIuJXEbFDvU2TJElSHVq9EfRFwBczc9PM3AT4cjVOkiRJa5lWA+DGmbnixs+ZeTWwcT1NkiRJUp1aDYCLImLFHW0j4lBgUT1NkiRJUp1avQr4JGBKRFxSDT8GHFNPkyRJklSnVq8CngXsHhEbAJGZ9v5JkiStpVoKgBExDHgfMBHIiLgR+FZmLqmzcZIkSep/rR4CvghYD7i8Gn438DrguBraJEmSpBq1GgD3zMxXdQ1ExI+Ae+tpkiRJkurU6lXAj0bEZk3DmwG/q6E9kiRJqlmrPYCLgVkR8Ytq+M3ADRHxTYDMfF8djZOk1fapjVqcb2G97ZCkAajVAHh19erykxraIkmSpDZo9TYwlwFExDrV8At1NkqSJEn1aekcwIjYISL+B5gL/CEifhURO9TbNEmSJNWh1YtALgK+mJmbZuYmwJercZIkSVrLtBoAN87Ma7oGMvNqYON6miRJkqQ6tRoAF0XEwV0DEXEo4OPgJEmS1kKtXgV8EjAlIi6uhucAx9TTJEmSJNWp1wAYEUOAt2Tm7hGxARCZae+fJEnSWqrXQ8CZuRw4tHr/jOFPkiRp7dbqOYC/iIizIuJvI2KrrletLZMkSVItWj0H8ITq32ObxiWwXf82R5IkSXVr9Ukg29bdEEmSJLXHKg8BR8S2EfH9iLgnIq6IiC3b1TBJkiTVo7dzAC8Gfgv8K7AQOK/2FkmSJKlWvQXATTPz9Mz8KXAysFNfVh4R+0fErIh4KCIujoi/OuQcERdExNyIWNrDtC9Uy/4uIo7sS21JkiT1rLcAuKTrTWYmsLzVFUfEUGAycFRmbg+sT883j74S2L2H5Q8A9qQROvcHzqvuQyhJkqQ10FsAnBARi7teTcNLquFV2QOYm5n3VcOTgcO7z5SZ0zPzyR6WPxy4NDOXZuYfgOnAAb3UlCRJUi9WeRVwZrZ6n8CejKbxyLguj1Xj+rL81b0tHxEn0zg8DcAWW2zRt1ZKkiQVZk0CXiuy6X3UsXxmXpiZ47peo0aNWo0ykiRJ5agzAM4BtmkaHg3MbePykiRJ6kGdAfAOYOuIGFcNHw9c04flrwEmRcTQ6rFzewM/6+c2SpIkFae2AJiZy4ATgasi4iHgeWBKRLw2Iq7rmi8ivhERc4Gh1e1gvlEt/3PgNhr3IbwROC0zn6mrvZIkSaVo9VnAqyUzrwfGdRt9B3BQ0zzvX8XyHwU+Wk/rJEmSylRrAJSk/jb29KktzTd7nZobIklrsbqvApYkSdIAYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMj4KTJA0un9qoxfkW1tsOaQCzB1CSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpML4KDhJWouMPX1qS/PNPufgmlsiaW1mAJTq1MozSX0eqSSpzTwELEmSVBgDoCRJUmEMgJIkSYUxAEqSJBXGAChJklQYA6AkSVJhDICSJEmFMQBKkiQVxgAoSZJUGAOgJElSYXwUnCTVwccAShrA7AGUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTC+CxgSRqMfBaxpFWwB1CSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgrjs4AlSVL/8TnUawV7ACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKoyPglO9fCSQVJ5WvvewWt/9sadP7XWe2ev0ebVScewBlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwtQbAiNg/ImZFxEMRcXFE/NVtZ1Y2T0TsFxHPRMSM6nVVnW2VJEkqRW0BMCKGApOBozJze2B94Jg+znNbZk6oXkfW1VZJkqSS1NkDuAcwNzPvq4YnA4evxjySJEnqR3UGwNHAnKbhx6pxfZln9+rw700RcWBPRSLi5OoQ8qyImDV//vz+aLskSdKgVfej4LLpffRxnjuBV2TmooiYAFwXEXtl5qMvWTjzQuDCruFx48Y1r0+SJEnd1NkDOAfYpml4NDC31Xkyc1FmLqrezwB+BUyorbWSJEmFqDMA3gFsHRHjquHjgWtanScitoyIqN6PBl4HzKqxvZIkSUWoLQBm5jLgROCqiHgIeB6YEhGvjYjrVjVPtYojgHsjYgbwY+BjmflgXe2VJEkqRa3nAGbm9cC4bqPvAA7qZR4y86vAV+tsnyRJajLvyT0AAAzhSURBVM3Y06e2NN/sdWpuiPqFTwKRJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqzLBON2BtNvb0qb3OM/ucg9vQEkmSBrdW/uYCzF6n5oYMEvYASpIkFcYAKEmSVBgDoCRJUmE8B1CrxXMxJElae9kDKEmSVBgDoCRJUmE8BCxJallLt7/y1A9pwLMHUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKowBUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwgzrdAMGvU9t1OJ8C+tthyRJUsUeQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqjAFQkiSpMAZASZKkwhgAJUmSCmMAlCRJKozPAh7sfBaxJJXF3/udsxZ99vYASpIkFcYAKEmSVBgDoCRJUmEMgJIkSYXxIhBJktYCY0+f2tJ8s9epuSEaFOwBlCRJKowBUJIkqTAGQEmSpMJ4DuBaynNBJEnS6rIHUJIkqTAGQEmSpMJ4CFiSpP60Fj0PVuUyAEqS1KJWzr/23OvBabD97D0ELEmSVBgDoCRJUmEMgJIkSYUxAEqSJBXGAChJklQYA6AkSVJhar0NTETsD1wIjABuAt6XmUtbnScivgAcDiwHzsjMq+psr9YeLT8K75yDa26JJElrn9p6ACNiKDAZOCoztwfWB45pdZ6IOADYE9gJ2B84LyI2qKu9kiRJpajzEPAewNzMvK8ankyjN6/VeQ4HLs3MpZn5B2A6cECN7ZUkSSpCZGY9K444EviHzDy6Gn4V8J3M3LWVeSLix8BXMvPn1bTPA3/MzK90q3MycHLTqG2BR2rZqNUzCphvfesXVtv6ZdcvedutX3b9Tm97d6Mzc8OeJtT9KLjmdBmrMU+vy2fmhTTOIRyQImJWZo6zvvVLqm39suuXvO3WL7t+p7e9L+o8BDwH2KZpeDQwtw/ztLK8JEmS+qjOAHgHsHVEdCXh44Fr+jDPNcCkiBgaEVsBewM/q7G9kiRJRagtAGbmMuBE4KqIeAh4HpgSEa+NiOtWNU817efAbcBvgRuB0zLzmbraW6NOH562frn1S95267vvWd/6pdXuk9ouApEkSdLA5JNAJEmSCmMAlCRJKowBUJIkqTAGwH4WEUMi4jul1ba+9d33rW/9smpbv/P114QBsJ9l5nJgs4ho+2fbydrWt777vvWt775v/bVH3U8CKdXdwH9HxP8Dnu0amZnt+F9CJ2tb3/ru+9a3vvu+9dtbf7UYAOsxisZTS/ZqGpdAO3aGTta2vvXd961v/fbXL3nbrb+avA9gB0TEOzPzu6XVtr713fetb/2yalu/8/VXxgDYARFxZ2buVlpt61vffd/61i+rtvU7X39l1rqTFgeJKLS29a3vvm9965dV2/qdr98jA2BndLLbtdNdvta3fom1rW99933rDygGQEmSpMIYADtjWaG1rW99933rW7+s2tbvfP0eeRFIm0TEDpn5YIdqD8vMpR2o+++ZeVa761a11wW2Bx7JzGd7m7+G+nsDuwH3ZeYv21Bv58y8p+46LbRjd2AMsBT4bTv3+YjYFRhdDc7NzLvaVXtl/O7V/90bQPt+R/a/iBgODMnMFyNiFLA/cH9m3t+O+t3aUtTv3R7qd3T7+8r7ALbPL4Ft6lp5RIwHLqtq/BT4cGbOqyb/msaXojYRcUYPo0+JiCUAmXl2zfXPz8wPV+9fC/wAeBwYHRFHZ+b1NdefAfx9Zj4VEf8EnAj8BHhHROyVmZ+psz4wMyJ+B0wBLs/MR2uu9xIRsRfwNWAxMB64Gdg8Ip4Ejs3M/62x9j7AxcD/AnNonHA9OiI2BU7IzOl11a7q+93r4HePzu/7Hdv/IuII4BtARsRJwMeAh4A9I+LMum9E3Omffad/73Z6+9eUAbAfreQXMTR+IWxQc/kLgX8D7gA+ANwcEQdl5mzacwXSvwE/Bh5oGjcEGN6G2gD7Nr3/DHBcZv4iIl4DfBPYs+b6QzPzqer9JGDfzFwQESNp/ExqD4DAh4B3A7+JiFk0/iB+LzMX1lwb4ALgsMx8rApEp2fm/4mIf6Tx+b+txtpfAw7v3gtUteO/aATSOvnd+4tOfPc6ve93cv/7OPBqYH1gBrBLZj4cEVsAU6n/RsSd/tl3+vdup7d/jXgOYP/6JLAujV+8za9h1P+HYMPM/Flm/ikzPwt8BPh5ROxMe65Aeg2NX0IBfK46/LQgM8/qwKGozTPzFwCZeTcwsg01l0fE5tX754A/V++XAEPbUD8z85bM/ACwJXAucCDwWERc1Yb6wzLzsaoh9wKvqt5fCfxtzbXX6ekQYNWOdvzs/e79RSe+e53e9zu5/0VmPpmZvwcey8yHq9pP0P5bj5T4e7dZJ7Z/jdgD2L9mANf0dO5HRLy35trDI2LdzPwzQGZeFxELgB8C69Vcm8x8BDg0IiYB06re0HaeYLpNRHyTxi+9zSNiRGYurqa1oyfk48AvI+IbwG3A1Ij4PvAm4Oo21F/xyz4zl9A4FPGDiNgYOKoN9f8YESfTOAT6TqreqGg8IL3u3zPTI+Jy4Fs0DsEljcOxJwC31Fwb/O51+rvX6X2/k/tfcyfOR7pNa0cA6fTPvtO/dzu9/WvEANi//gl4ciXT3lRz7cnA64BpXSMy81cR8TbgSzXXXiEzL4uInwJfaVfNymlN76fT+MO7OCK2BK6tu3hm/jgi7gFOonES8J+BccAFmXlD3fVZyTZm5gLgojbUfy+NnpcP0Dj08s/V+FHAv9Zc+0TgPTTOf+o6Cf8x4PvApTXXhoH13fsJ8J/tqlnp6HdvZTXauO93cv/7VESsl5nPZeZPukZGxA605zm0pf/e7fS+v0a8CliSJKkw9gD2s4iYCBxO0+0AaBwWvnEw17a+9TtdfyVtOqPuq2Ctb/3qfM9tgFuqnseu8Ydn5jVtqv2rzJzfztrW77E992Tmzu2uuzrsAexHEfFZYD8at4R4jOp2AMCxwI2ZeeZgrF3V/wyN+091qr7bX/D2r6Jdj2Vmbbdfsr71I+KDwCnA74DXAh/MzB9U0+7MzNpuA9TJ2taH6vZD3Y0FZgNkZt0XwK0RA2A/iogHgR0zc3m38cNo3Jhzh8FY2/rW7/C+39MvYWiE0Fdk5oi6alvf+hExE3h9Zj4XEdsD1wBfy8yvR8RdmbnrYKxtfYiIqcDTwGeBF2nsczcDewNkm+9J2VceAu5fy4FNadwQtNnLqf+qvE7Wtr71O1l/Y+DvgQXdxgftuQrY+mXXH5KZzwFk5kPVqRA/isaNoOve9ztZu/j6mXlwRLyLxo3Az8zMaRGxZKAHvy4GwP7178AdEXE9f7kdwCuAicBHB3Ft61u/k/VvAkZl5szuEyKiHY/jsn7Z9RdG06M+M3N+RPw9jauA6z4XrJO1rd+o+Z2I+BlwXkQcx1qUqzwE3M+qe08dyF9OhJ8D/DQzFw3m2ta3fqfrS50QEbsAz2e3515Xpz+8PWt8HFsna1u/x/a8BdgnM1f2VLABxQAoSZJUGB8F148iYklELG56vRgRcyPi29U5CYOytvWt775vfeu771u/vfXXlAGwH2Xm8Mwc0fQaSeM5nXcDXx+sta1vffd961vffd/67a2/pjwE3CbRhkvSB2Jt61vffd/61i+rtvU7X78V9gC2QUQMAYaWVtv61nfft771y6pt/c7Xb9Vac7ny2iAiXt/D6JcD7wZ+MVhrW9/67vvWt37765e87dZfcx4C7kcRcUO3UQnMA64HLsrMpYOxtvWt775vfeu3v37J2279NWcAbJOIGJOZc0qrbX3ru+9b3/pl1bZ+5+u3wnMAaxQR60fEcdF4OsJvSqltfeu771vf+u771h/gMtNXP75ohOq3AN+l8VzURTSejjBsMNe2vvXd961vffd967e3/hq1vdMNGEwv4DxgLjAVeBewHvDIYK9tfeu771vf+u771m9v/TVuf6cbMJhewFPAdOBIYEQ17uHBXtv61nfft7713fet3976a/ryHMD+tSXwReAdwGMRMRlYp4Da1re++771re++b/321l8znU6gg/UFbAy8D7gZ+CPw1RJqW9/67vvWt777vvXbW391Xt4Gpg0iYlvg3Zn56ZJqW9/67vvWt777vvUHJgOgJElSYTwHUJIkqTAGQEmSpMIYACVJkgpjAJQkSSqMAVCSJKkwBkBJkqTCGAAlSZIKYwCUJEkqzP8HmysvlBK6WHEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 750x525 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Set our x axis \n",
    "x1 = happy_au_arr.index\n",
    "\n",
    "# Get our y values, which are proportions\n",
    "y1 = happy_au_arr.values / happy_au_arr.sum()\n",
    "y2 = nervous_au_arr.values / nervous_au_arr.sum()\n",
    "\n",
    "# Set axis arguments to control output\n",
    "ind = np.arange(happy_au_arr.index.size)\n",
    "width = 0.3\n",
    "\n",
    "# Plot figure, bars, ticks and legend\n",
    "plt.figure(num=1, figsize=(10,7),dpi=75)\n",
    "p1 = plt.bar(ind,y1, width = width, label = 'Happy')\n",
    "plt.xticks(rotation = 90)\n",
    "p2 = plt.bar(ind+width,y2, width = width, label = 'Nervous')\n",
    "plt.ylabel('Proportion')\n",
    "plt.title('Action Unit Occurence By Smile Type')\n",
    "plt.xticks(ind + width / 2, x1)\n",
    "plt.legend(loc='best')\n",
    "plt.savefig('action-unit-occurence-by-smile.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "for  AU01_r (1.0235940321165866, 0.30602706519804435)\n",
      "for  AU02_r (-0.5638279704202742, 0.572871214265066)\n",
      "for  AU04_r (-1.009143077489369, 0.3129060211649882)\n",
      "for  AU05_r (0.5563454734615338, 0.5779747063736267)\n",
      "for  AU06_r (1.2276797146323757, 0.21956722432193032)\n",
      "for  AU07_r (1.0916023234904817, 0.27500793442653637)\n",
      "for  AU09_r (nan, nan)\n",
      "for  AU10_r (-0.5084982247269069, 0.6111039837774244)\n",
      "for  AU12_r (3.2466237765125094, 0.0011678265425233078)\n",
      "for  AU14_r (0.5497549326621741, 0.5824874735787529)\n",
      "for  AU15_r (-1.1448438976048245, 0.25227382739316817)\n",
      "for  AU17_r (-1.4784400203689474, 0.1392900401651598)\n",
      "for  AU20_r (-0.06584204669242268, 0.9475035804251846)\n",
      "for  AU23_r (-0.31111352959521155, 0.755714316069206)\n",
      "for  AU25_r (-0.7301598446278202, 0.465292484503251)\n",
      "for  AU26_r (-4.596325629444343, 4.300065088491656e-06)\n",
      "for  AU45_r (-0.8667583376026822, 0.38607443392072516)\n"
     ]
    }
   ],
   "source": [
    "# performs 2 proportion z test between happy and nervous data for every AU\n",
    "# Output is (Z-score,p-value) for a 2 tailed test.\n",
    "for kk,_ in enumerate(nervous_au_arr.index):\n",
    "    count = np.array([happy_au_arr[kk], nervous_au_arr[kk]])\n",
    "    nobs = np.array([happy_au_arr.sum(), nervous_au_arr.sum()])\n",
    "    print('for {}'.format(happy_au_arr.index[kk]),proportions_ztest(count,nobs))"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}