Explore_Subtitles_vs_Predictions.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import webvtt\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib as mpl\n",
    "import seaborn as sns\n",
    "\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Loading CSV with the timestamps and prediction probabilities"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "filename = 'Barack Obama\\'s final speech as president – video highlights.webm_clipsize1000_predictions.csv'\n",
    "\n",
    "df = pd.read_csv(filename)\n",
    "\n",
    "df['timestamp'] = pd.to_timedelta(df['timestamp'])\n",
    "\n",
    "df = df.sort_values(by='timestamp').reset_index().drop('index', axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABIUAAAFICAYAAADOAk5gAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzs3Xm8XVV9///3hwyAhEEggIYhiGANaB0Q9ddWbdUW1IKzoNahVtqvpbU//VpxqK1grVNt61f8VtA6VBEBp1hDUVvQqoUSIUBCCAlISAiZE0Lm6fP94/PZ7p3jublnuveec/br+Xjcx7rnnH3WXmvtvfbw2WvvY+4uAAAAAAAA1MsBE10AAAAAAAAAjD+CQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAGBBm9jdm9pUOv/smM/vJfj6/zsze2GxaM9tsZo/rZL5tlvFgM/uumT1sZteM9fyazN/N7PHjPd9OjbZMB52Zvc7Mvj/R5QAAYJgRFAIAYAyZ2f1mti0DK6vM7ItmNm2iy9XI3c9x9y+N8Nk0d79PkrL8HxqjYrxS0rGSjnL3V400UQZD3MxeM0blmDBmNtXM1hbriJn9npn92MweMbM1ZvYjMzt3osvZC2a2IPvFZjPbY2bbK6/f6+5fdfffnehyAgAwzAgKAQAw9n7f3adJepqkMyW9v3ECC3XfL58k6R533z3KdG+UtF7SG8a+SOPuOZLmuftmM3ulpGskfVnS8YqA2Qck/f54FsjMJo9Fvu5+egYcp0n6L0kXFa/d/cNjMU8AALCvuh98AgAwbtz9QUnXSTpDkszsRjP7WzP7qaStkh5nZo81s9lmtt7MlpjZWxuyOcjMvp4jR241s18vPjCzi83s3vzsLjN7WcN3zcw+nbdn3W1mz698cKOZ/VGzche3VZnZhZJeJ+kvczTHd83sXWb2jYbpP2Vm/zRCXk/MeW3MkSLn5vsfVAQ8XpN5v2WE758k6bmSLpT0e2Z2XOWz55nZcjN7b462ud/MXlf5/Itm9s9m9oNsox9lfs3m82Izu83MNpnZMjP7m8b5NEx/v5m9IP8/y8zm5ndXmdknK9M9y8x+lvW/3cye1zDrF0maY2Ym6ZOSLnX3z7n7w+6+191/5O77rBNm9gkz22BmvzCzcyrvH25mnzezh8zsQTP7kJlNqnz+h2a2ML97fbUtcpn/qZktlrS4SftcZ2YXNbx3u5m9PAOc/2Bmq7MN7jSzM5q18/7Yr97G6Gb2NjNbnMvvUjM7Jdtzk5ldbWZTK9O/xMzmZVv/zMye3G4ZAAAYdgSFAAAYJ2Z2guKk/7bK23+gCHAcKmmppKskLZf0WMXtVB82s9+pTH+eYvTIkZKulPRtM5uSn90r6bckHS7pg5K+YmaPqXz3mTnN0ZL+WtI3zezIVsvv7pdL+qqkj+Vojt+X9BVJZ5vZEVnHyZLOV4xuaaz/FEnflfR9ScdI+jNJXzWzJ7j7X0v6sKSvZ96fH6EYb5A0192/IWmhIkhVdVzWb4ZiRNHlZvaEyuevk3RpTjMv69PMlpzXEZJeLOl/mdlLR5i20T9J+id3P0zSKZKuzvrPkPQ9SR9SLL//LekbZja98t0X5TRPkHSCpGtHmdczJS3K+nxM0uczoCRJX5S0W9LjJT1V0u9K+qMsy3mS3ivp5ZKmK0bqfK0h75dm/rOazPdrki4oXpjZLMVIr+/lfJ4j6TTFuvhqSetGqUerfk/S0yU9S9JfSrpc0usVbXVGUSYze6qkf5H0x5KOkvRZSbPN7MAelQMAgKFAUAgAgLH3bTPbKOknkn6kCH4UvujuC/KWqeMk/Yakd7v7dnefJ+lz2vc2qZ+7+7XuvksxkuQgxQmy3P0ad1+RI0q+rhjhcVblu6sl/aO778rPFykCHh1z94ck/VhS8QygsyWtdfefN5n8WZKmSfqIu+909/+U9G+qBBda8AZFMEyZNruF7K/cfYe7/0gRpHh15bPvufuP3X2HpPdJenYG6xrrdaO735lteYciCPLcFsu4S9Ljzexod9/s7jfl+6+XNMfd52S+P5A0VxEIkpmdImmyuy9SBDIk6aFR5rXU3a9w9z2SviTpMZKONbNjM9+/cPct7r5a0j8oAnaS9CeS/s7dF+a692FJT2kYOfV37r7e3bc1me+3GqZ/naRvZrvuUgQ5f02S5TxGq0erPubum9x9gaT5kr7v7ve5+8OKUXhPzekulPRZd7/Z3ffk87J2KPsKAAAIBIUAABh7L3X3I9z9JHd/W8NJ9rLK/4+VtN7dH6m8t1Qx6uVXpnf3vSpHFcnM3lC5XWajYuTE0ZXvPuju3pD3Y7uqWfiSIuChTP91hOkeK2lZlrtahhkjTL8PM/sNSScrRlNJERR6kpk9pTLZBnff0pB/tY7V9tuseDbRr7SBmT3TzG6weLjzw4ogytGN043gLYpRMneb2S1m9pJ8/yRJryqWTy6j31QEcqQI4lyX/xcja6ojvZpZWanP1vx3Ws5riqSHKvP6rGKEVlGWf6p8tl6SaYR1rVGuo99TGWS6QDnqKoN9n5Z0maTVZna5mR02Sj1atary/7Ymr4uHuJ8k6Z0NbX2CerO+AwAwNAgKAQAwsapBmhWSjjSzQyvvnSjpwcrrX45qsXgw9fGSVuSIjSskXaT49a4jFCMprPLdGZVbi4q8V3RR3sK3JT05nxvzEo18S9YKSSfYvg/Ubqzf/rxRUZ95ZrZS0s2V9wuPNrNDGvKv1rHaftMUt3E1a4MrJc2WdIK7Hy7pn1W25RZJj6rkM0lxC5Ykyd0Xu/sFigDMRyVdm2VaJulfM0BY/B3i7h/Jr75I0pz8f1FO/4r9Nch+LFOMjDm6Mq/D3P30yud/3FCWg939Z5U8mi3rqq9JusDMnq0YsXZDpQ0+5e5PV9x6dpqkd3VYj04tk/S3DfV7lLs33iIHAECtERQCAKBPuPsyST+T9HdmdlA+GPctiuf2FJ6eD/OdLOkvFCf+N0k6RHESv0aSzOzNygdaVxwj6c/NbIqZvUrSE1UGIVq1StLjGsq9XfHsmysl/Y+7PzDCd29WPFD7L7MMz1P8ktZVI0z/S2Z2kOI2sAslPaXy92eSXmv7/kLWBy1+2v23FEGqayqfvcjMfjMfSHyppJuy3Rsdqhi1td3MzpL02spn9yge+P3ifE7S+yX98lk1ZvZ6M5ueI6I25tt7Fcvx9y1+Zn5SLuPnmdnxZvYoxa1+N0hSjuh6h6S/MrM3m9lhZnZAlv3y0dorb9f6vqS/r3z3FDMrboH7Z0nvMbPTs8yH5zrRjjmKETmXKJ4FtTfzekaOtJqiCKBtz/qPpysk/UmWw8zskFxeh476TQAAaoSgEAAA/eUCSTMVo1e+Jemv3f2Hlc+/I+k1kjYoHlL98nxG0F2S/l7SfysCN0+S9NOGvG+WdKqktZL+VtIr3b3dBwB/XtKsvCXn25X3v5TzHOnWMbn7TkUQ6Jwsw2ckvcHd725hvi9V3B70ZXdfWfwpHiY8WfEsIylup9qgaL+vSvqThvyvVDxke73igcWvV3Nvk3SJmT2i+FW0qyv1eDg//5xilNMWxW18hbMlLTCzzYqHTp/v7tsy+FQ84HmNYjTLuxTHY78j6b8zwFbM51rFsv7DrM8qxUOqv9NCe0nxvKWpku7KNrlWeTuau39LMYrpKjPbpBhVds4I+TSVzw/6pqQXqHzOkyQdpgjKbFDcvrdO0sfbybtb7j5X0lsVt7FtkLRE0pvGswwAAAwC2/fRAgAAAO0zsxMl3S3pOHffNEFleJ6kr7j78SN8/kVJy939/eNZrlaY2WckzXf3z0x0WQAAQH1MHn0SAACAkeUzgt4h6aqJCggNgXmSvjvRhQAAAPVCUAgAAHQsH6C8SnGb0NmjTI4RuPuozwkCAADoNW4fAwAAAAAAqCEeNA0AAAAAAFBDBIUAAAAAAABqaMKeKXT00Uf7zJkzJ2r2AAAAAAAAQ+fnP//5Wnef3sq0ExYUmjlzpubOnTtRswcAAAAAABg6Zra01Wm5fQwAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaaikoZGZnm9kiM1tiZhePMM2rzewuM1tgZlf2tpgAAKCXZl78vYkuAgAAACbY5NEmMLNJki6T9EJJyyXdYmaz3f2uyjSnSnqPpN9w9w1mdsxYFRgAAAAAAADda2Wk0FmSlrj7fe6+U9JVks5rmOatki5z9w2S5O6re1tMAAAAAAAA9FIrQaEZkpZVXi/P96pOk3Samf3UzG4ys7N7VUAAAAAAAAD03qi3j7WRz6mSnifpeEk/NrMnufvG6kRmdqGkCyXpxBNP7NGsAQAAAAAA0K5WRgo9KOmEyuvj872q5ZJmu/sud/+FpHsUQaJ9uPvl7n6mu585ffr0TssMAAAAAACALrUSFLpF0qlmdrKZTZV0vqTZDdN8WzFKSGZ2tOJ2svt6WE4AAAAAAAD00KhBIXffLekiSddLWijpandfYGaXmNm5Odn1ktaZ2V2SbpD0LndfN1aFBgAA9TTz4u9NdBEAAACGRkvPFHL3OZLmNLz3gcr/Lukd+QcAAAAAAIA+18rtYwAAAAAAABgyBIUAAAAAAABqiKAQAADoKzw3CMOA9RgAMAgICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAAD6Brfgjh+CQsCQYMMJAAAAAGgHQSEAAEBgGQOB9RQAgN7q66DQeOz4ObiYOLQ9AAAAAGAkY3HOyHnovvo6KITBUXQsOlh/YDkA6Fa/b0f6vXy9NhH17Yc25vgCANAO9hftIyiUWHkAAMCwaPW4pnG6To6HOIYCAGBwERQCxslIB83tHkxz8A0AAAAA6AWCQgD6GkEwAAAAABgbBIVGUdcT0rrWezyMZ9uyHAG0i+0GGvVqpCsAAOg/fRkU4iADY411DAAGC9vt3hq2ZweNVrZ+LjsAABOpL4NC/YCDBwyKdtbViVyv6VMA0D62nb1FewLAcGL73rm+CgqxIIcHyxIAhkOz7Tnb+P7DMulfLBsAaN14/ggP2+fQV0Gh8cCCryeW+8Sh7QG0athuaULnWK4AAIyP2gWF0L/qdgDYbX3r1l4AgMFWl/1WXeoJAOheP+wzBjIo1A8Nh8CyAAB0in1I7w1Smw5SWccbbQMA+yq2i40pujcQQSEW+HBgOfYPlgUAoGo89wv9uA/qRZn4BTQAwCAaiKAQBls/HQT1U1nGU13rDaA/sU3qnYloy35Yfv1QBgAAhgFBIaCi04PMZsMYOWDtLdoTQF0M2/Zu2OoDAMAwmdCgEAcJQP3Upd/XpZ4AMJGGbVs7bPUBAPS/gR4pxI5z4tD2Y6fuz3Wou14uE5YvMHza7deDsB2o6y1wAIDxwTZ//wYqKDQeC5MVBnVEIKL/NLYj7YpCr9eFYVm3BukYYVjavNGw1gsAgGE2UEGhVnFQ0ntj2aYsr+HEcgUADDL2YwCAOuiLoBA7XQCDhG0W+kEdbrmhr9VLHdbpfpk3ANTBIG1nJ7KsfREU6tYgLWyML9aNztBuAMYC2xagOyP1IfoWAIyOZ7c2NxRBIaBqkDpgnRXLiecZAQDqrJN9F/s7AECvEBQC+kQ/HOD1QxlGMwhlBABgJO3sx9jnAcDgGNRtNkEhjLvROkunnalfO2G/lguDgfUHvdaP6xQjBsce7QIAAJqpRVCIAyG0i3Um8LPoGGusUwCAYcZ+DmgNfWXiDFVQiOG4Y68f260fy4TxU7flP6z1HdZ6YfAMyro4bKNqAQAYNGPxjNSJ0FJQyMzONrNFZrbEzC5u8vmbzGyNmc3Lvz/qfVHbN+gLB+OHdQXDhF+nqZexuiW37viFEgyCiVh3WF8BYLiMGhQys0mSLpN0jqRZki4ws1lNJv26uz8l/z7X43ICwJgbiwNdDp4BAACA/lX34/VWRgqdJWmJu9/n7jslXSXpvLEtFjA+6r4BQHd6sf6wDgIA0P/YX2N/WD+GWzfLdxDWjVaCQjMkLau8Xp7vNXqFmd1hZtea2QnNMjKzC81srpnNXbNmTQfFBYDhMAg7CAAAhkE/7HP7oQy90A/16IcydGPQy98rtEP/6NWDpr8raaa7P1nSDyR9qdlE7n65u5/p7mdOnz69R7PuL6zcwPigr40f2hrYP/oI0FvD0KeGoQ4TpZ/brp/LBnSqlaDQg5KqI3+Oz/d+yd3XufuOfPk5SU/vTfH6x7BsAAaxHv1Q5n4oA4Dx0U/9vZ/Kgt5j+Q4HluPgGZRlNijlnGi00+hoI+xPK0GhWySdamYnm9lUSedLml2dwMweU3l5rqSFvSsi0D8mcoPaTxtzHsg82GjriTNsPyPer+XqV7TXYOqHnxweq3kPY53G2qCWu1Xj+Qumw7BeD9v6MGz1QWtGDQq5+25JF0m6XhHsudrdF5jZJWZ2bk7252a2wMxul/Tnkt40VgUGMP7YQYyun9uon8uG4TJI61o/nOg36qeyAGgP/XfstBqoGsuA1iAFxYB2tfRMIXef4+6nufsp7v63+d4H3H12/v8edz/d3X/d3X/b3e8ey0Kjd0bbmLKxAnqLPtWasWynfgwGNOrXA1gA9K1+VPdl0mr9695O+FWsE4OvF8uwVw+aBnqOjRSA8dQP25xB/cnTfmi7fjQIAchhQ1t3bhjbrp2LnRNR/2Fsc/S3dtc51tF9jUd7TESbExTqEAd6wPCqa78erd790C6tlKHd4eP9UK/xUJd6Ahhfg75tGYTy9+P+a5DbDeNrGO9EGYY6VBEUAvZj2Dr8IOinhwMO2k6s38vXjUGv26CXH61hOddLPy7vfioTt8Gi11gfBls/BjcnQj/Wl6DQOOr1CtCPKxTqhXUQI+nVujGot1NhuLFutYZ2AgYX/bdeur2tjPVlsBEUGiBEV/sPbT8YBuG2qH5CewD7ok+0h/bqf4O2jPppFHFdytCuQSxz1SA/GqSTMg9iPQddP7c5QaE+1umKw4YBQL+q27aGB5cCE4v+0H+GZZmM53F6L747Vka7aN2PZZ4o473O9LLtWY69N5YXrdv9LkEhDKy6bJwGoZ6DUMbxMkxtMSh1GZRy1lk/LKN+KAPGD8t77NHGzQ1KuwzCQ6qH9ZeeRtJtu/RTXTBYCAoBwDhgR917g9ym3Is/cWhroH/RP1tDUKAeWJ4YLwSFhhgbEqA36Eut66erf2Np0Ms/1mgfoD/U9RaeutV3LHX7AOLx+u4gG+QfI+Ii13AgKDQBRtpB13XHjeZYD9CqQVhXBnEI+CC0KwD02iBs+wahjNgXywzoXwSFxlgrG8CJOFliw4w6Yr1vjnYBgBLbRFTxC6ZjjzZEv+mndXI8RuETFGpTP/+STD+tvGNh2OsHjDf6VO/Rpv1jEEenAcNoUH9OflDLjcHXD+sDz62qF4JCA4jOCAyf8bxnv5f6pRyDatjbb9jrB6B1+9se9MNPbgNAXREUAjB0OEhEFetDvTHaFugMfWL88Wy8/tNtwJJlMNjqsvwICo2RuqxAANBLbDsBYOyM5TaW7TeGST+uz/1YJkycXq4PBIUADC3uh554tDXqiPUew4j1enCx7HqPY8z9ox3Gzli0LUGhHmHFB1AXdd3etVPvurYRAAAABgtBIQAAAGCIEJiuNy5ijK6u9cbgGst1lqAQAAD7wYEjAGDYsa8D6ougEAAAAICe4GHW6GesQ8CvIiiEgdMPG/N+KMOwo40HS92XV93rDwAAMOjqejxHUAgAgA7V9eABAAAAw4GgEAAAAAAAQA0RFAIAAAAAYIAwWhm9QlAIAAAAAACghggKAQAAAGiK0QgAMNwICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAABQQwSFAAAAAAAAaoigEAAAAAAAQA0RFAIAAAAAAKghgkIAAAAAAAA11FJQyMzONrNFZrbEzC7ez3SvMDM3szN7V0QAAAAAAAD02qhBITObJOkySedImiXpAjOb1WS6QyW9XdLNvS4kAAAAAAAAequVkUJnSVri7ve5+05JV0k6r8l0l0r6qKTtPSwfAAAAAAAAxkArQaEZkpZVXi/P937JzJ4m6QR3/14PywYAAAAAAIAx0vWDps3sAEmflPTOFqa90MzmmtncNWvWdDtrAAAAAAAAdKiVoNCDkk6ovD4+3yscKukMSTea2f2SniVpdrOHTbv75e5+prufOX369M5LDQAAAAAAgK60EhS6RdKpZnaymU2VdL6k2cWH7v6wux/t7jPdfaakmySd6+5zx6TEAAAAAAAA6NqoQSF33y3pIknXS1oo6Wp3X2Bml5jZuWNdQAAAAAAAAPTe5FYmcvc5kuY0vPeBEaZ9XvfFAgAAAAAAwFjq+kHTAAAAAAAAGDwEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAABQQwSFAAAAAAAAaoigEAAAAAAAQA0RFAIAAAAAAKghgkIAAAAAAAA1RFAIAAAAAACghggKAQAAAAAA1BBBIQAAAAAAgBoiKAQAAAAAAFBDBIUAAAAAAABqiKAQAAAAAABADREUAgAAAAAAqCGCQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAABQQwSFAAAAAAAAaqiloJCZnW1mi8xsiZld3OTzPzGzO81snpn9xMxm9b6oAAAAAAAA6JVRg0JmNknSZZLOkTRL0gVNgj5XuvuT3P0pkj4m6ZM9LykAAAAAAAB6ppWRQmdJWuLu97n7TklXSTqvOoG7b6q8PESS966IAAAAAAAA6LXJLUwzQ9Kyyuvlkp7ZOJGZ/amkd0iaKul3mmVkZhdKulCSTjzxRFm7pQUAAAAAAEBP9OxB0+5+mbufIundkt4/wjSXu/uZ7n7m9OnTezVrAAAAAAAAtKmVoNCDkk6ovD4+3xvJVZJe2k2hAAAAAAAAMLZaCQrdIulUMzvZzKZKOl/S7OoEZnZq5eWLJS3uXREBAAAAAADQa6M+U8jdd5vZRZKulzRJ0r+4+wIzu0TSXHefLekiM3uBpF2SNkh641gWGgAAAAAAAN1p5UHTcvc5kuY0vPeByv9v73G5AAAAAAAAMIZ69qBpAAAAAAAADA6CQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAABQQwSFAAAAAAAAaoigEAAAAAAAQA0RFAIAAAAAAKghgkIAAAAAAAA1RFAIAAAAAACghggKAQAAAAAA1BBBIQAAAAAAgBoiKAQAAAAAAFBDBIUAAAAAAABqiKAQAAAAAABADREUAgAAAAAAqCGCQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACooZaCQmZ2tpktMrMlZnZxk8/fYWZ3mdkdZvYfZnZS74sKAAAAAACAXhk1KGRmkyRdJukcSbMkXWBmsxomu03Sme7+ZEnXSvpYrwsKAAAAAACA3mllpNBZkpa4+33uvlPSVZLOq07g7je4+9Z8eZOk43tbTAAAAAAAAPRSK0GhGZKWVV4vz/dG8hZJ13VTKAAAAAAAAIytyb3MzMxeL+lMSc8d4fMLJV0oSSeeeKKslzMHAAAAAABAy1oZKfSgpBMqr4/P9/ZhZi+Q9D5J57r7jmYZufvl7n6mu585ffr0TsoLAAAAAACAHmglKHSLpFPN7GQzmyrpfEmzqxOY2VMlfVYREFrd+2ICAAAAAACgl0YNCrn7bkkXSbpe0kJJV7v7AjO7xMzOzck+LmmapGvMbJ6ZzR4hOwAAAAAAAPSBlp4p5O5zJM1peO8Dlf9f0ONyAQAAAAAAYAy1cvsYAAAAAAAAhgxBIQAAAAAAgBoiKAQAAAAAAFBDBIUAAAAAAABqiKAQAAAAAABADREUAgAAAAAAqCGCQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAABQQwSFAAAAAAAAaoigEAAAAAAAQA0RFAIAAAAAAKghgkIAAAAAAAA1RFAIAAAAAACghggKAQAAAAAA1BBBIQAAAAAAgBoiKAQAAAAAAFBDBIUAAAAAAABqiKAQAAAAAABADREUAgAAAAAAqCGCQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGmopKGRmZ5vZIjNbYmYXN/n8OWZ2q5ntNrNX9r6YAAAAAAAA6KVRg0JmNknSZZLOkTRL0gVmNqthsgckvUnSlb0uIAAAAAAAAHpvcgvTnCVpibvfJ0lmdpWk8yTdVUzg7vfnZ3vHoIwAAAAAAADosVZuH5shaVnl9fJ8r21mdqGZzTWzuWvWrOkkCwAAAAAAAPTAuD5o2t0vd/cz3f3M6dOnj+esAQAAAAAAUNFKUOhBSSdUXh+f7wEAAAAAAGBAtRIUukXSqWZ2splNlXS+pNljWywAAAAAAACMpVGDQu6+W9JFkq6XtFDS1e6+wMwuMbNzJcnMnmFmyyW9StJnzWzBWBYaAAAAAAAA3Wnl18fk7nMkzWl47wOV/29R3FYGAAAAAACAATCuD5oGAAAAAABAfyAoBAAAAAA6eMpEAAAgAElEQVQAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaIigEAAAAAABQQwSFAAAAAAAAaoigEAAAAAAAQA0RFAIAAAAAAKghgkIAAAAAAAA1RFAIAAAAAACghggKAQAAAAAA1BBBIQAAAAAAgBoiKAQAAAAAAFBDBIUAAAAAAABqiKAQAAAAAABADREUAgAAAAAAqCGCQgAAAAAAADVEUAgAAAAAAKCGCAoBAAAAAADUEEEhAAAAAACAGiIoBAAAAAAAUEMEhQAAAAAAAGqIoBAAAAAAAEANERQCAAAAAACoIYJCAAAAAAAANURQCAAAAAAAoIYICgEAAAAAANQQQSEAAAAAAIAaaikoZGZnm9kiM1tiZhc3+fxAM/t6fn6zmc3sdUEBAAAAAADQO6MGhcxskqTLJJ0jaZakC8xsVsNkb5G0wd0fL+kfJH201wUFAAAAAABA77QyUugsSUvc/T533ynpKknnNUxznqQv5f/XSnq+mVnvigkAAAAAAIBeaiUoNEPSssrr5fle02ncfbekhyUd1YsCAgAAAAAAoPfM3fc/gdkrJZ3t7n+Ur/9A0jPd/aLKNPNzmuX5+t6cZm1DXhdKujBfPkHSOklrJR3dYqo2pu11yryZN/Nm3sybeTNv5s28mTfzZt7Mm3kzb+bd7/M+xN2nqxXuvt8/Sc+WdH3l9XskvadhmuslPTv/n5wFsRbynttO2sl3epUyb+bNvJk382bezJt5M2/mzbyZN/Nm3sybeQ/CvFv9a+X2sVsknWpmJ5vZVEnnS5rdMM1sSW/M/18p6T89SwMAAAAAAID+M3m0Cdx9t5ldpBgNNEnSv7j7AjO7RBGBmi3p85L+1cyWSFqvCBwBAAAAAACgT40aFJIkd58jaU7Dex+o/L9d0qs6mP/lbaadfKdXKfNm3sybeTNv5s28mTfzZt7Mm3kzb+bNvJn3IMy7JaM+aBoAAAAAAADDp5VnCgEAAAAAAGDIEBQCAAAAAACooZaeKdQrZvZrks6TNCPf2ilpgaT7JG1191vMbJakt0u6W9JZklZJ2iJppqRZkhZJ+jtJfynpDElrJX1D0lfcfdO4VQYAAAAAAGCAjdszhczs3ZIukHSVpOWSflfSOZIOVQSndkh6RNJRkvbke15JXdI6SSZpmqRfSJoiabekXTnd29z9xnGpUDKzY9x9dcN7R7n7uvx/sqS3SHqZpMfmJA8qHtx9iKKu/0fxi20vVwTDXubup+b3Hyfp/ZJWSPqIpH9UtNskSRsUdW85vxHy/AdJz5a0UNK73P3+TuperXcnep3faHl2u2wyj67bcizqDSCY2bEqL0Q86O6rKp9Nc/fNZnaku683s3PdfbaZHaEYSXukpKdKWuDud5nZTEm/odg+bJa0XbHNuFWxvZCk0xQXOn5TcdHjFZJ+rthWP1bSMyXd7u7XmtnvS/otSQ9IWq3Yn6lHeW5QXES5v4f13qT4hdFtreQp6cej5LdUse8+sof1tmzPB0fJb56k10n6tOICVC/ybGd535btqB61ZavrZDv1fkDSLZIWu/t67YeZPc3db83/D5N0qqT73H1DvneKpDWSDlYb/bGFere1Tra4ni/Nvy29qre7bzKzIyWpR3lK0uGttuME9u8HJT261XK2mGfby6aXbTlEZRyEZbNGOYChZtugseiL8yRN7WFbShO4Dcr3erlvHG3/PVPSqxXxiimZ9uqYoOXjtTbqvVTSPHffrBaNZ1DoHkmnu/uufH2n4uT5JsUCO1bRUe+TtFfSgZKOVwR/9igWwOGSDlM0wh8qTt6vVjTISyR9x92fOko59hckOFzSeyS9VNIxOclexWiklZKWSHq8pJslvU3Sf0j6HUnPyf8966Is83WZ1+R8vVjSFZn3hYog2GMqn3vW7QCVgbDdlTIcKOmgnHZHfu8ixcrQmN9uxYFsEVjbmu27MOtQzXODpKOzfSdlPrsq9V4g6fSGeq+RdLakR2V9i7So9zkNbbE9/xrbcrWkF+T707I8+8tvb5Ztt2KkWbPl02qenS6bXYr1cb2kIyQtk/RtxS/wbR6lLTut9+4s49clfVfd7SSmSXpWluv4rFuvTm7bOSHbLOmPWsizrZPbFuttmfZqh9vOCVmr9e7ViehMDdaOrFc78BdJ+r+KvrUyv/8ERb/crjj4OCTb+HRJGxX99gOSPpx5TqrMe5VimzBJ0Wd2KPrv9GzXvZlvsd2dkn/bFf16T362Peu/ozLN5Hz/YUl3ZTt0mueB+eeKbeTOfH1AF/U+TuUFmt35nUObtOV2xYUdKfrWYZUyNuY3Kd/fqTh4+YWk53ZR74Oynnty3jsV29bJDfmtU5wQbM46fF7SWyv17iTPdpd30b5bFOvzrC7astV1sp16T816F/Mq5rtFsexXKPYdGxQjtrdkPsUJplfKOF/Rz4vHFexRGQT99S7q3eo62cl6XpRzW7bj1i7rXRxob1AcExzXgzylWB9XK445elHvXvdvKdbB1YqLwU/qMs9Wls0dkl6kuIh3iqQ3SPqaymXQSVsOWxkHbdnsUWy3DmqS57Bug3rdFz0/25xtdmyP2nK8tkEP5jTVPB/OfIvBJJ3uG0fbf0vSnYp1ZUvOb3P+P0XdHROMdrzWSb2LZb4rp50j6e3u/rBG4+7j8qfYCJyk2CjckZUt0u2KA/yiURZL+k428K2SvqA4mZ6X6V7FLWPrFAf7OxQn6Ltzod8t6UZJn8v06vz8HsXKvy7/tuXfZkmvV3SQvZnPpmzMmxUr/BaVgZr1Od26fP1ILpAHM31E0kOZzypJ75D01azvvYrAzIpKG+zN9NZsh+JgcV1+dr/ixGe3YsX4R0kfUxxMPqS44tiYX9HhLV///X7yvFnSTyT9qyLAsT4/K+q9pEm99+bfnsyjsd4LJP2bpOtzeTyU7zW2ZRHc2a1yw9Msvwszn7mK9eO7ipOIq7rIs9Nl80XFOrciP79BZUcdrS07qfcn8vvbc57LVZ6gbMrXG/I7xUbl4cp89uR7OxQ74K2Zl2eZHsmybs9pd+V7mzMt3t/SkOeWLMfGSn678zvrcn5Ffqszv6UqN/jrmpSzmuc2levYtpx2e5f13qZyO7Mty9VJW+7MNiuW397Mp7EdO6n3aG3Z6rLZogi6b1NsF4tt1cou8mx12SzMdrxd5Xb53ko5OmnLdutdHPDuzToX/Wdt5lH0xX/L738nXy/Lz+/J+RU71r1Zh/Pys7VZ16XZnpsrea5WrENrVQauVym2B9sr7fc0xfb1B5nHt3K5fKWLPBcpAt7zc/qHsk6ruqj3QklvVrmvWaPY/zTm+XCWe5PK9e3hJvnNUmy/v5v1+I+sQzf1/oUiWHhf5rFB5f6+mt8/ZVk+ma9/kXnO7yLPdpb3XYrA6EOS/lmxXfpsF23Z6jrZTr3vUFxw+h/FMce9Ktehap7bslzF93bk/zvyb42iXz4r5/c9xXHSD7Kc3dS71XWynfV8nqT3Svqp4hhpWda9m3ovkvSjrP81WbfPdpHnPZLerRjB9aBinVzYZb3Hon/fk8t5heIYfHOXy7vVZVMcp92hsm/uyPp02pbDVMZBWTZ3Ks5bblT0nVWVeg/zNmgs+uKdkn6Wy+YzimO4QdsGFefbP63kuVvlcXI3+8bR9t/zc7rfyXK+INvjU+r+mGC047VO6j1LMQr464q+9lZJ17YSqxnPB03/heLA7wmKk/qNipMBUxx8vFjRGLsUIyl+Oz87PSt4siJKd5KicZ+liIwerAgMvFLRQD9UXIF+huLWoOcqbv2ZJulxisjhEYoo21RFw/5Q0pcVC/lERYRuqiLyd2JOb4qFKUnnStru7kcpFta9ipOfZYoFsFLl1cCDss6vzfQERYR8euY1M/M+WTFKZXK+vkfRIW7Psvxa5jct6/kuxVW/o7KNGvO7qpKXJD1/P3meLukSRWDs0ZXPDs7vF+vJbZJ2ZL135rK6XRGoaaz3yZJmK24TnJHlnlnJs2jLexWdYXG25Uj5XZDLwRXrxyG5bI7tIs9Ol81bFevNDMU6M0cR0T60hbbspN6vUazzaxU7yWMU6+uGzHNGts2jVJ4wH6Yykr0tp/uJ4la4Vyp2jm/KeXxB0R9WKfrljlxeh+R3N+TfwSoDqg+rvNp0mKLv/pbiBGKnYp1ZV8nvoMyvCP7+S37vUfvJ8wFF312Y+WxV7Ci6qfcixbZot6T35ecPdZDntpx2qWKDviQ/b2zHTuo9Wlu2umx+oRg58+Jstxcq1seru8iz1WVzTLZjMWz1p4ptd/FeJ23Zbr1XuPsxKq94bVQcCB6a7fpvWba7JO109/NUHmwtVhzs3JvLbkNOO1OxLp2iWDc2KvpqUaeP5XRTFP34nnxd3C70UM5vRb4/W+V2/AjF6EDLcnaa5+MUz9wrRp+Y4srT3i7qfZpi/Smuqu7JdmzMc5tiOd6beS5WBNQb8/uvrPMMxbL5eubbTb2Pz++cpFg2j2T5GvM7QdIud39HlnFr5rm8izzbWd4nK9blYxV9ZorKCwydtOVMtbZOtlPvJyr2i89QLOcjVF4N3ZPTSnHgu1hxXKEs188zXaxyVOaViv58nOIY7CbFNrCbere6Traznj9JcaX8/1O5nT6qy3pPV3n8syXbcmoXeZ6cbfn0rPcOxVX6fuvfx2XbHafY7x6k7pZ3q8vm7xXHqU/O/IrR7iu6aMthKuOgLJtfy7o+N+s/WfuO3h/WbdBY9MWTFMv4pKzzoRq8bdDsnG6qpL2Z506Vd3p0s28cbf/9tnz/y5nnj7Mc89X9McFox2ud1Pu/FOc5p0qSu1+hODcd3XiNFMrRQgdkBd+nOLF+kaSDK59fmelvKE6Gr1AMU/v1fO8cxS0CpyuGHa5SRMP2SPrPXDg3ZLpVZYRtUTbcQpWjj27KBbRYETl1xQn7tTnd1kwfo+gIP8yyFVfF9ypuNduT094j6W8UC7y4Gr1VMapnZeZfXEnfreh0p2eexVX2h7MMrljZ3ivpMkWH26mIxs9XeUV95X7y+6nKlWaPojOMlOemrONeSf8r670o61HUe2u+t1cRuS/aYKR635vz3qaIUi9UnOxV8ywipdsVV4yLDVuz/D6Yy/OB/N6pWYfjusiz02XzOMVG7S8UkfeHso3/sKEt/2SEtmxWxpHa8YOZ/k9+59Is41/ntOsqeaxUrNu3qVzPb1AEnOYpgkuu8upCMVTzwmzXFYqdzCpJH83PNmR+P6vk+e+KDVCxMdqbdf6ZyhFzxclGkd97VUa7dyl2TJuzzUfLsxjxsVJl9L3jeucy8GzLlYoTs3bzvF/lSJmvZbq8STt2U++R2rLVZfOcnK4YWTa1YXl3kmery+YLmd8tioMCKdbjHV20Zbv1fkTl7W3FCNQ3Zf6fzjJtUCx/V+zo9yr610rFM1fWK3b0xf7kzny9Q+XBU9FX35t5rsx8PNujKPOXFf2uGL22ShGwXKXyCt2H8vNLu8hzRea1Q7HdKLa7m7qo9+L8zvZsgyVZjsY8ixGkxZXgkfK7NpfvVpXbuW7rfb/Kq5rzVY6Eaczvvsxnqcr1rts821nexfb3EUl/nJ+/qYu2bHWdbKfea1Re0bxBsX9amvlenXkWB8O7cxrPZfpRxbZgg2K/uUnxbL5lKgPIb+5BvVtdJ9tZz5epHFV4S4/qXWwbtyoudu5UBP07zfP2/O72LN/iHtR7LPr3Q1nmLYoLtN0u71aXzUaVF2CKE6tu23KYyjgoy+Yelccudyr24w80yXPYtkFj0RfXZhk2qxzlMqjboOJC+lqVI//H+nhtg2K/uERxnFmcP/+1uj8maPV4rZ16X6uIcdyquFtniqRFrcRpxu2ZQmPBzL6vGOXzZsXondmKne6tku5y9xeY2TZF1G2n4irkSxSNe6XigKw4YXu54oTkyYpo6mGKlXO14krPE9x9mZm9NN+/WrEQjlYMeX+hogMcrghcFRuzUxUn35sUwbBXKSK2H1J5K8VrslynKaKy0xVXEbfktJMUw49PV6xo0xSd9lOS3inpKe5+UqVd/iznfYXiCuxLFRvqp6gcvWCK5+A8R3GCd4JixSvub9yp6FDnN9T7q9leL8jZrd9PvacqDjyL5x8VgbZqnq9QDO/bkvVe2ZBfcQtJtR33qHy+0kM5z07yPK2S53sVO4dDst02Zd6vVaxTMxQblBmZ12OyXqYISN6sONk+UHElsIjkHjRCWzYr4+r8brN23Kbo7J+Q9FeKkWKfkvRld7/IzDaofODbfMWVm1WKjcNrs1zrFKPo5is2Ek/I8l0q6Y0qr/B8yN0/bGYrFSM+LMt3nGKH+l3FOnWQYv0tbic6VhH4XZzteGiT/FYqgrxbVV6pGCnP4krBVMVO6MjMb1oX9T4sl920XF7vy7Ie0maexQiXFYog4WTFDnTSCO3YTr1bbcvRls0WxbpX7HAen9P/rWJd7yTPdpfN2pzvuqzHQT1oy3bqvVqxTZfi4GWZ4mrmSe6+1eIh1Acqnq8yTXG15gLFvuG+rNeZiu3tLsWB1NEqDziOVhxofVPSbHffLklmdobiYPsYxailbyn69KMzv+/kfJ+X87hN5dW2kxTbyZ91keepKg8gjso2v1nSR7uo92NU3tb3aMX+4scNeT5H0a+KPG/cT36HZ72XKbZ7pyiChI31fpIi4N5KvY/P5fELRR9aqwgYVvM7QnEwV+T3QsWVwG7ybHd5369yBM5xin5+ZxdtOT3Ltb918gjFhYfpLdT7txVX/g9W+byKlYr18iKPZ62dqbhQ90aFExXby6VZxlmKbcNWxYijg7MNVyj66cz8vJt6t7JOtrOeP0dxLOSKY4BpimOc27usd3E78sG5fBYrAuwj5XmC4uRjpDwPzWW8JP/fpjj477f+vUTlKM/jFVfIO13erS6bPYrb9Ip2/CvFaIFu2nKYyjhIy+ZwRb9Zr3L09i3qvN8MwjZorPriSsUx2LTM8z7tfxs02nZtorZBByoGNlynePzHIYpgylger31TETO4VNLvKbbntyr23zPU/TFBq8drlynuTHlnC/U+PNeJD+Qye6K736RRDHpQ6NGSLlbcqnOIotGKh6a9xt2XtxgkWKLYOP1fxQLcoTh4mpuzepniYOhuxZDGuxUHSfcpTq7+VXGb1PcVC/yJOf0TFZ3nWSqvWh+e836G4sRcipO/KYpI+Cn53UNVRoNPVKwwazLP5YoDyR2KDdwZihX9cJUPoHpY0QHWZjon57VRsSI/VXFA/lTFRvQ3FcGNpypO1s7KeRcPzKqm8xUd97nZpsUtalb5zolZ3qOzHs9WnOQVQ+anVz5fl+8XDxYrPl+n6CBrFRvwo1TucG5UHFh/LZfPt5qkr1ZsxGaN8PlHFCegh+Ry86zba7KsxRBGV3TSw7MuxcPBVitOYlflMjxOEWz6jGKU0FFZlzXZLgc0pMU8iqu2RRs3Tlec4D4ml2txUtHujuzfFc+TuCyXx1rFMp+s2Jl2e3LbysnT4Ypfa/t3lQGZH+8nz1Y2lu3We5OiD0/Ncna7wy362Xzt/4TsfW3Uu5W2HO1kudiRfUixfZrIHVmrO/BWTm6flGUsTm5Hq/c9kt7s7o+Y2cGSjnH3pUJXih9t2E96rLuvamG6X/kFzx6W8Sh3Xzda2u602JeZmaRD3X3TRJdlPI1Fvevalr2W7TjN3R+Z6LKMpK7Lmn7T3wah72BsDHRQaH/M7M3u/oVqqnjOzjvd/UMjfH6o4pawNYoTwWWKk7ONipOeBxTBjv9R3Eu5WnEiUjzJ/ADF7R4nVqbdrjjx/7YiSPGQyqe0b1NEroshZjMUJ1ZHK4IEUzLP+xUBh4NUPrDqIJXPqSkCC9VgwoKsw10qn1hePBNAKn99qXhiuqkcxTAp2+AYRaDqRJUBkj352YmKE72XqAyWLVKMPinS4v2HFCe2ixWR+SIw1U66VHECuFvlLyhNVRkk2txBOkXS/1aMenpxtmvxrKCdKoN2y7O+xSiFdYoT0iI4tDTrN1URJDswy3VMvj5Y5VDCIoAm/WqwrVm6J8s6WfGAvKdLepu73yj0xDCd3LZz8jsW5ewnVv6a5MsU/Xeq9u1/Va5yOPYB+be/aYth79+S9BF332hm17n7OR2k31eMEH2j4iroMzLdptjm3byf9CzF/qj6/lzFhYgfZrpdcTGheIZTsX3ZqwiIPjpfb1Rsa2Y01L34ztrKd4uLERcoAnCvbUgvUATsG9Pi8xcrRu4+UzHM+RmKWwqOVQQLNykuvLSTFsHToxXB88fm62NVXmBYl/XdqXIE5lZFsHKTYj9bnXa6YhtefH6wYgj4sxTHE+e3mV6ZeX8s532EYsRq8by80fYHzdK9uZwmqXz+n48wfbFPKY4vTOWvYzVOq8r0xTxG6hOt2KVYLvMVx0Of9/LXaC939wvbSRW3ut+oGEX9EsVzLIr0ufn+c3Oa4xT94nyVfex/skzVaX+iGCVVjDrfmO8XzwJslyuO34q2X6B45uCt7n6amd3Tbqp4Nt61lfR6xXHVYxTHpMdkeojKY0JTjMA7XHFM2zjtqVnOmYpj0FmKY44piuU/qc10r6Kv7Mo8liqOpz+nOA6+RjFqvkivHuX9qxWjCN+p8nlzT1f08cZno+5Vecy+R7HO7ci6Npt2h8pRr1J5+38nfbF6jFdNm/WbXSofGDs5227Kfqb9lb7TSb/J9ApFf/hjxXb4hYr9xU7t2x+apc9RXNDqpN+4ynMjU6x76xTnCY31btZ3Xubup3bSbzK9391nmtn9inX+k2red1pNW+03exXrV3HOtUSx3zpS5XN9imkPVLlO7Fbsvz+quKDZSb/5tmIf+H6N3nda1U6/2aXyR00mK85dJ6t53yme21o8W6fZOtHqcVr1O2N5vLZNsc7foLE5XhvtOK26v1blO2sU688Biousb3f3h0ZpK43rM4XG80/SA+2m2XDLMl2hOFj8cDbwyny9MhfiLsVV912KDcHenHZrw7RbFUGRrblizlf5ZP0i3aZy9E3x+cJMi9d35edH5byKp44vyc//f0WHvzunuyXL9pWc7/cVwY0dintrt6l8js5yxcHz4nx9b+YzP8t9Us7zpPzezyvzKKZpTBdX0m2Vci1UeRtZy2kuo2JDsEMxJNGz7MWzV9pN91Ze787/V2db764sg2q6W7EhdJU/lbhK5UNI78p2n5913qzyvtZ5HaT/rbhn+EGVv9RX3TC2+7dH5UZ6m2KnvVAxauqIbOfr2k0VO7h7FSc+qxrSXyhG0/2npL9T+euCixUHEVcpRo4U0xbpNYr++IUs421Z/w2KDWMr6Yas472ZblAE167I5fN9RV9/brZxY/qcEdLi8+L2izdlWX+o8hfs7lb5qz33Kg6Mr1E8cH9+ZZrGtLh/eIPigGNnfrZHsX62mu5S+dOixWizXYqDsJ2KA/T5ioOFdtPzFDvBOYqTjSKgWhzo7lT03+Kgd3PWpzpNY7pT5W2XxYFgsc62m+5WOcquCKw+rFjf7m34u1SxbszJMs7J15c2mbY4mP6aoq9fpxiJ9GnlPeBtpu9T+TDBXYr90B7Fdn2dYn1vN92g8hlFOxUHBzuzDbZX0mI7d4fKnxrfodhvNU5bLItq2ou/XZk+kmW5SuWvALabPlJJ16kcpblU5a+ObFKMHN6T6VZFUKp4dk7jtKu076/07c10T37Wbroh2/qzir69RuXI5fUqb5VoJ12RZVqi6IcP5mebFNuRIp2f7bJUsZ7dl9N9t8m0G1U+z2FRlrF41kEnf3MU+7KrFCeQX1EEM56qGHX4UAfpt3L53ZbtuzLTFYp1qUhvULk/39zQfo3TFtvNYkTwxny9XeUxUztp8UyN6jZtLP525fzur6TFsy8W5HL8eZbr602m3V5J16jsU8Wv5bSbvl/RV+Zn+VarPF7b22G6W3HMXTyHY53KbdWOSnp7tndR32IZNpu22oceyby35N/KDtKtKp8vUuzLF6h5f1it6FvrMl2d7zdO+wNF3/mmou9creg7z1Nn/eZpWd7Z2QbFfrlZf2g1bbXfrFH5LJUtKp+vsrLJtMUx61j1mZ2Zf7P+0Graar95f7bT+xXrxn9nO3yvybRFmxSBl2raSb95JMv2zypvnRup77SattpvHlT5q533K47BViu2xY3Tbs/8timOKdcrRrJ3cpzW6+O192nk47WNuWzH6nitneO06nceUZxnXS/pzyR9p6XYyUQHb7oM/NyRf9vU3Uak8TtFPsWJvucKUTwg8v8odjpF+smc9t8r0xTpykyLDUbxwMG5md6s2IndpggCFPds3l75vDjwmZf13qoIRG1VeRLxA8UGoLjt5Av5+X9l23w8p9lYmfYElQ/MWqnYgRcP1LxNZYCnCEQ9OvO6rZKOFBT6TuZ1TZbnh5n+LNuh3fTlik74ckXHOC3TcxUbhoc7SG+tLOttKh/gvSrTorPvzToVV1jvVnlb32qV608RmHPF+lBcDb5HHQTCMp2b7bY803dnur8N4Uh/1Y3q7YpnEt2huOLa7cnt/KzzYpU/vV7dUP4sl9+iLH8x3f5Ofls9uR0trW5UvbKci7SXf0UApDg4+L7KoOwmlSeg+zv5bfXkdrS0evK7Q7HDdcWBYbcnt8WyW5t/jSe238h5fDWX4cbKMuz25Ha0tDj53aUY1bFNcUXzPklfUnkQvkj58D2VD8NuTBc1/O1R2cd7caJXBEU2ZHpDpj9VudPvJt2rWO93KrbXv0yLQHtln/JAJb2t4TvvzLw+nu388crrrR2kT1L8Co0U6+Y8xfa4SLd1kC6spFsa0mYXGCYr+uJNlbRxG3xrJb1X5dW3YjvUbtp4cWJ7tnezNm81vb2S3qbyeKE4MS3SW1Q+CHa7chRJZflXp93ekN5TSdv+y3m4yoddF/fSmekAAA8mSURBVOt9L/+KoPB6xfamSD+XnxXpo3IZfqvJtOsq6d5s012K9bKT1HKeX848i/QRxQlLJ+mdipO74jlkxbFj9bhsm8oR7FPy9bxK2jjtokq6tdu0YduyTdIfZDsUJ7xbukz3ZjuuVnfbtTsq6W2Zf9F3Ojlea9wOLc60WX8YaZ/T2H/2ZBl73V+KPviLTJv1h1bTVvtNdWTINsU+qkgbp/1U5nWFYv94RbZFp/3mWP3qPmekvtNq2lK/aVgnq9vgW5p8p1j/bless5tUrgOd9JfiwvkylcenS9TdPqfVfnNz1rNID9S+fXN/fecelT/U0+5xWq+P13Y1pNXjteKi0Zgcr7W5Pat+p0jnVdPR/jodOtYvjlX8CtlmRbBgo+Kh0w+rfFJ6K+krFQt4ocpfttmguNJwgWJBHaJo9EMUw+8OUAwZPEAxhPBAlVHE6rQ/zFSZWk57rmKo2EtUBhuKk8rnVdLi9qw9kjxvibhbsTN8puKhsbsUHf9IxTOW1iuGFf+jYgTI57KsyxVD/Ypp3604wSueUzNL5U+on6Hyp9TnZblvVwz9O6OSFtM0pk9XHHQ/M+v7/Kz/sztMv6EYmXOt4iDnCsUor9mKk+DHtJu6+9MUD8H9oKJDPqJYl45QBGOmKAIKaxXrQBEJPznzuF+xLhRP5v9g1nV7Lr93KtaP4rlDT+gg3ak4iP1Y5vUvkg5w979SOQqjpb/8zjR3f5Hi4btnKE7OPizpbMV6/KeKkVDtpqdn3U9WrCvFrY3FKJD/zPJfk589S2UAbqSRI8XGdotiXSqGPi5W+RDmUVN3n5Xpk1UGHlYr+uYnshyfUOyM2k0XKtaD6s5ySf4d5e6/K2lykaq8BfTXimmapMsr6UGK9e9Rih3B5FZTd/9GLoPv5DwfUKybr1YcvBUnLId0kO5V3JJRDIEublVYKknu/opMX6cyqL5b5c+oqklaXFXfovJBg0tVDq9uNd2W5dmqWK9vVjxc+oHM71BF354saYqZXSppi5nNkbQ10y1mdonKoeBFuk5x5XeVYv+wShHwXKDy53VbTt19SrbLUZk+P9PPZf23d5Ae0ZCelm1xgGIfWaTK9pJiXfhyJd1nWnf/e8V+Zaai/zw+l9XjK6/bSS/Jtn+PymeaPa6STukgvVuxDfpB5lncYnxE1vEISTKzu3I5rs3lV7TPCY3TKoZxr1Bsi5eoHM30gMpfUGsnfUvO+zLFvmabYtkfmPNXB+ljK2U+WrFMj876H1lJ16scWu6KHytwM3t3k2m3K26d267oL+sztU7+zOz8zPsTmb4905NU/spMu+kixTaySBcr9uEHqfyRhz2KZahK+oNsg3lNpj2wkv6/9s4uxq6qiuO/PW2hkNoCU0rTtLSlVAkmmho/KFAeJBCDobEkIg9UaGJiNNgYDIYoLyZq/IgQXvQBjQFMISatWlAD+IAYAqEPVlvlo2kL1La0DNCPKdOZucz24b93z57Tc2fOOffc9g6z/snJv52771737M+11957rRHk4/H98L3jVdlLGz8KPIja4qHArZB3Hb4ctZlFqH8uQvW9COlIkdeG33Ez6g/zE86n3RF+19/Cdw6T1V+rKjvn/gSMOOc2kV2ZGUT6wVto7KzKB5E+9i6ZvnwBaqd9kZ1z96A2983w97lB/px8WtTW+9D8mvb9uvPiaMhzNMh7J3Bhf3DOPZDjW/NpQ13EeWN7wrX6jffeIX1wVuDloe6K+kNZLtVvQn/4IDBkvk7/UpB2Y/j/ilCeK1Cd1+03jwEznXO3oDYf+1DkGTW4VL8J/eGIc24Pmc/UE2RBf9I+tBzNOUvQmPZuyL9uv/lhqIuHQt04dPWoqD+U5bL9ZjgpD4+cNHvn3NqCtCfJfPpeFMqgn3p6WqP6WqKnXcnp+toepCt1Q197L5TtRHraewVMwn05nhAzJ0/S03gSNeytaPD4A9rN28x4vz4Tsvd+i3NuLzLqrAc2Iiv1RlTxW9Fi/J/AKu/9E865LWhn8aqE96O76qtyaX+FjAyfjuy9PwAccM7NR3cTox+D17z3w865qwPnP58F3OG9HwAGnHMHUSO/FA38zwPHvPd/RkcTieyc+2KSJk37veTza5L/n48Mb4eK2Hu/d7I0qHF/KpTjxUh5q3o/PfJSpBSc471/wTn30fB+X/FybFuJAbz3e4AfhDvWEXORweRdNDi9GX53/nOHOvK/kFX7y6ENvIGMYleQnaKZjTp8Hf524JeQ0fOlZCAsfc89N6gOobb4WRRx4I+o78wj86dRhaPPqsGEF6CB7zjy6+ITfi6U5a/RsdbjSdrIF5AZ6SZa3E7GkA2q0WB6J6rXThe3y0O+8W74ZWRtdpZz7juh7O8mW/hFjj4Eiha3H0OLl6/TZnE7GecWv/HkI6Fc4uL2KnQk/dqK/DlkuLkS1X80Xi4Jsh8L/NsgL/ozm2jxmy5u+2i/uJ2M30Dtfheqnxi5cA26/ueQcSPi+6gN3BDK6caQ5j6KERWCePXox2j8XUB1vwSg6zufz/E+NO+sRrtRVfgpVD+R493159BJ0v8Bi51zlwO7nHOfQLvFv0ebKz8h8z2Spn0W+Wh6OJTfB8jIPlaTh9H1yxYaQ88h20wpPX8nvB8paCtD3peF8p2H6jfyMjKjc/TzcC7ZZk+a9hyyaHfxOP4G4FHv/Z3OucerMIBz7gZk5N8Rft9yNFfG8XpeRY79ZUnyO6OfwfPCu56H2sZh1BdBwTVGUTt3ubRxIR8V4EtCnnExVxWbAv885LkW9ZsL0ebVqbZWgZdQ7FtqFZlfrs+gU9Or0fyZcjylnaZ9gfE63bbAcRd/XxV2zq1AOssRstPf8Rr8rJocN/Hi/H+CbGOxRbYpdXv493qyK8Txmlk+7YxQdmtQm786yErnxyp8c/ht8bTqSmCnl3+717331znnnqvI16MgHsNIf+0PMhaFMo38U7JTzGPImDlKFu1vLPcdj8aK2HcuRXrCYjR/VuHDZEbVBWSGu3y/2RT+dlco+7tCuscpxoLwDrPRGHdf+M11+g3AdufcF1KmuD+U5VL9JvSHAadgEUeBB4Cb0FrrNwVpd6CAH0+h+oP6/Sb6+fllyOck2WnpMcb3h7Jctt/cTNYWWyHNTNQeXS5t1Jf6kPGmD23kPVGn3wA453aF9347yOunfX8oy0X95j3UxmL/mRnKeFl47y2h7LagMTFNewjNBwNow3FNePf5ob6q6mkRTehroBtBfRTra4Nog71xfa2KnsZ4fe0fgV+LPEk5AXx4HU0bDB8WuCzK3jqkCE/kMHcyeDjlNDQqhofQ5PsjZHirs7hdRrEz7zhQfg0pQj8DvpvwrcjwsRopFin/nWKHbX+lmhJ0D1pg34bCM96LBtRH0Omuj6PrIamT9LI8iCatEaSwbWP8wnYzUowj70GT/39ov/jdjE647AKuCXUVF2qlDIAJx8XvHKQQzEVG5ItDGf/Oe39bWLRW4U+GOoynIPML2+jbJzq/HUaT8oUT/NaoELkkj2Gyqz6tkjyCFM5jSGnZS+aLIRom47HkWJdzkDHuS+iq342BL2pT76+QRaLslD/SRnY+kmUTXEX2RL97D2qbz3fAK1Efv5PTI3g2VbZ1yvxMyH4V+eV6Bim5zyAFvZP6LSN7ThvZRe+9KtTP9WjsuAUplueRBcmownGTBXS68gAag+rkVZUH0Jg3GxmNq8quu6Fzkixox7lojB9EG4Sd5JnyO2QnSfOcvvebyHB6dZu0+fdtssybqu/0N8cToScK+B00L6f1fU2btKnOcnyCPMvyAFmk411o3v8Gim5cxA+ik3NMkOZUWu/97c65R7z3X22KAZrOswI74OHJ0qKbDTvRhmzH7L3v79J7P+q9Xz8ZB9ml0napvp9ExvmtHfAGZHTfAPwCRZi9v4DvIAu40N8mzd3A/Ul9P4Q2rdvpi9Ef70Rp8jxWMt1kHN2kVJHdFNeRvdN7/zRl4HvAN5A99thT70GhtjvmJvOaSrKRsnlfUzxV3ns6yEa7jwfRlb4WWpiMkp2cej3wi+Hz/WiyHUXGqNEkTbf4xSCnSPbekKZbXEX2VHjvKrLL1nc3yryFDNZno62Vlb2PzG/LCDISxNMoJ2twi8yh7clE7rEO8jTZJrvXZI8hg1C6oeHJrrtFPlnA8cRKPm3TbLJNdi/KHgvPSPhbC+lvH6BN1EE0bw2SBVkaQe4RYpqJ+L8dcLzq3U52J3l3Q/Zgws+gTbh7S60pz+QC1h577Gn2oUaUvSJuMi+TbbJ7QTbdiybZJJtsk92rsg+SBfB4GSmcr9TkpUjxX4gWBscSrpunyTbZvSZ7J+ovu8kCRoyhE9UjCccgNdsSPowMSfm0bzfMRbJjVLCmZXUiO18OnXI33rsT2Vbfp3/nKFmkwd2oX8WTMUtRX1qK5qUyUbCb4qkme2nCO9Apxh1l1pSlHA8ZDIazB+fcv8MzFJ6x8HhgSQUea8c18jLZJrunZaNj4ovR9cCF6Grat9Bx2nj94Hx0XbEP7ab0oauOI8gX2lAubdNssk12L8qO12EG0QK7HynyQ3XYe/8G4L33b5EtmiPXytNkm+xek40CeLTQ9dohZEwd4/Sowlcgo+um8L1nw2dDBWm3N8xFstPIdN2QWUd2vhw65W68dyeyrb7Hp52N5psjyIA0FJ6IwcDHAkf7xZnwjTzVZA8ChLGw5b0/gcalSWFGIYOh93EJzUbZ21DAVfMy2Sa712W/TPPRJJtmk22ye1E2CYMMqTHC0KtV2Slq6pBzbiHqmzcFvrZunibbZPeabOQvbCe6ujwLnf7oY3zk3zQC8Ao0Ny1rk6ZbbLJNdq/JBgWguJLmomA3xVNN9nZghpOzduecixEQJ8WZsHIZDIbO0HSUvd3Ix0XKlfKqyibbZJ8F2Y1Hk2yaTbbJ7lHZT5NES014L/WiU85C0WRagUeBNd77E07hievkabJNdi/KvgNdhbkOLdzeR9Fd85F/x0UAdsXRgbvNJttk95LsU8EofLNRsLvCU0T2OrQRuo4SsOhjBoPBYDAYDAaDwWAwGAzTEHZ9zGAwGAwGg8FgMBgMBoNhGsKMQgaDwWAwGAwGg8FgMBgM0xBmFDIYDAaDwWAwGAwGg8FgmIYwo5DBYDAYDAaDwWAwGAwGwzSEGYUMBoPBYDAYDAaDwWAwGKYh/g+ZREbenK+CywAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1440x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df = df.sort_values(by='timestamp')\n",
    "df['timestamp_sec'] = df['timestamp'].dt.total_seconds()\n",
    "\n",
    "df['prediction'].plot(x='timestamp_sec', kind='bar', figsize=(20,5))\n",
    "plt.title('Probability of Applause/Cheer vs Time');"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create new columns for peer-based relative percentile (e.g. probability at timestamp X:XX is within the 75th percentile of probabilities)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "start = 0.6\n",
    "end = 1\n",
    "for i in np.linspace(start, end, num=1+int((end-start)/0.05))[:-1]:\n",
    "    name = f'{round(i, 2):.2f}'\n",
    "\n",
    "    df[name] = df['prediction'].apply(lambda x: 1 if x >= df['prediction'].quantile(round(i,2)) else 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "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>timestamp_sec</th>\n",
       "      <th>0.75</th>\n",
       "      <th>prediction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.533033</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.550475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1.5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.554168</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2.0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.568925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>2.5</td>\n",
       "      <td>1</td>\n",
       "      <td>0.540498</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   timestamp_sec  0.75  prediction\n",
       "0            0.0     1    0.533033\n",
       "2            1.0     1    0.550475\n",
       "3            1.5     1    0.554168\n",
       "4            2.0     1    0.568925\n",
       "5            2.5     1    0.540498"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['0.75']==1][['timestamp_sec', '0.75', 'prediction']].head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### TODO - Find sentences said before applause"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "start = []\n",
    "end = []\n",
    "text = []\n",
    "for caption in webvtt.read('Barack Obama\\'s final speech as president – video highlights.en.vtt'):\n",
    "    start.append(caption.start)\n",
    "    end.append(caption.end)\n",
    "    text.append(caption.text)\n",
    "subtitle_df = pd.DataFrame({'start': start, 'end': end, 'text': text})\n",
    "subtitle_df['start'] = pd.to_timedelta(subtitle_df['start'])\n",
    "subtitle_df['end'] = pd.to_timedelta(subtitle_df['end'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "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>start</th>\n",
       "      <th>end</th>\n",
       "      <th>text</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00:00:00.560000</td>\n",
       "      <td>00:00:10.740000</td>\n",
       "      <td>\\nhello Scavo it's good to be home</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>00:00:10.740000</td>\n",
       "      <td>00:00:10.750000</td>\n",
       "      <td>\\n</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00:00:10.750000</td>\n",
       "      <td>00:00:14.360000</td>\n",
       "      <td>\\nif I had told you eight years ago that</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>00:00:14.360000</td>\n",
       "      <td>00:00:14.370000</td>\n",
       "      <td>if I had told you eight years ago that\\n</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>00:00:14.370000</td>\n",
       "      <td>00:00:17.540000</td>\n",
       "      <td>if I had told you eight years ago that\\nAmeric...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            start             end  \\\n",
       "0 00:00:00.560000 00:00:10.740000   \n",
       "1 00:00:10.740000 00:00:10.750000   \n",
       "2 00:00:10.750000 00:00:14.360000   \n",
       "3 00:00:14.360000 00:00:14.370000   \n",
       "4 00:00:14.370000 00:00:17.540000   \n",
       "\n",
       "                                                text  \n",
       "0                 \\nhello Scavo it's good to be home  \n",
       "1                                                \\n   \n",
       "2           \\nif I had told you eight years ago that  \n",
       "3          if I had told you eight years ago that\\n   \n",
       "4  if I had told you eight years ago that\\nAmeric...  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "subtitle_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If the subtitle is auto-generated by youtube, there is a lot of overlapping of text between timestamps."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tf2-alpha",
   "language": "python",
   "name": "tf2-gpu"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}