Lambda-School-Labs/Labs26-StorySquad-DS-TeamB

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Documentation notes on this notebook:\n",
    "\n",
    "##### Purpose\n",
    "- Used as an exploratory notebook to look at both API and human transcriptions, potential Squad Score formulas, and correlations with story rankings and textstat packages\n",
    "- Training data: 167 stories\n",
    "- Generate a pickled scaler from training data for use in production\n",
    "\n",
    "##### Outcomes\n",
    "- Ultimately, API transcriptions used over human transcriptions to generate scaler and formula for production, as it's closest to what will actually be seen in production.\n",
    "- MinMaxScaler was selected over StandardScaler to maintain positive score values\n",
    "- More features are explored in this notebook than were included in the Squad Score formula\n",
    "- Initial Squad Score includes only features generated with Python & Pandas that represent either features used in validated writing complexity metrics OR features requested by stakeholder, AND are versions of the feature that are impacted as minimally as possible by inevitable errors in spelling, handwriting, and transcription. \n",
    "- Weights are initialized at 1 for each feature.\n",
    "- This Squad Score results in a -.63 correlation coefficient with provided stakeholder rankings.\n",
    "   - 1.0 features: story_length, avg_word_len, quotes_num, unique_words_num (-.60 correlation)\n",
    "   - 1.1 features: story_length, avg_word_len, quotes_num, unique_words_num, adj_num (-.63 correlation)\n",
    "\n",
    "##### Future Considerations\n",
    "- Scaler should be re-fit with any newly provided training data\n",
    "- Major room for growth in this formula is to improve the weights to something other than 1s. This choice was made due to lack of labels, and reluctance to create an over-fit formula, but if different techniques are available or labels are provided, the weights should be adjusted accordingly to improve the formula."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import glob\n",
    "\n",
    "import joblib\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import nltk\n",
    "import seaborn as sns\n",
    "import textstat\n",
    "\n",
    "from sklearn.metrics import mean_absolute_error as mae\n",
    "from sklearn.metrics import mean_squared_error as mse\n",
    "from sklearn.preprocessing import MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Uncomment to download nltk modules\n",
    "# nltk.download('punkt')\n",
    "# nltk.download('averaged_perceptron_tagger')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Import API-transcribed stories\n",
    "path = \"../../api_transcribed_stories.csv\"\n",
    "\n",
    "api_t = pd.read_csv(\n",
    "            path, \n",
    "            usecols=[1,2],\n",
    "            header=0,\n",
    "            names=[\"story_id\", \"transcription\"],\n",
    "            dtype=str\n",
    "            )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compile human-transcribed stories for comparison\n",
    "\n",
    "root = \"../../Stories Dataset/Transcribed Stories/\"\n",
    "\n",
    "transcriptions = []\n",
    "\n",
    "for file in glob.glob(root + '**/**/Story*[3000-5999]*'):\n",
    "    with open((file), 'r') as file:\n",
    "        story_id = file.name[-4:]\n",
    "        transcription = file.read().replace('\\n', ' ')\n",
    "        transcriptions.append((story_id, transcription))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generate dataframe of transcriptions\n",
    "cols = ['story_id', 'transcription']\n",
    "human_t = pd.DataFrame(transcriptions, columns=cols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calc_metrics(df):\n",
    "    \"\"\" \n",
    "    Cleans transcriptions and adds various metrics as columns.\n",
    "    \n",
    "    Features included:\n",
    "    - Grade level\n",
    "    - Length of story (in characters)\n",
    "    - Average word length (in chars)\n",
    "    - Number of quotation marks\n",
    "    - Number of unique words\n",
    "    - Number of adjectives\n",
    "    - Percentage of complex words\n",
    "    \n",
    "    Note that code generating proportional/ratioed features is commented out, as correlation\n",
    "    to rankings sharply decreased with this change in feature quality.\n",
    "    \"\"\"\n",
    "    # Avoid SettingWithCopy error\n",
    "    df = df.copy()\n",
    "    \n",
    "    # Clean potential transcription errors\n",
    "\n",
    "    # Strip leading or tailing spaces and integers \n",
    "    df[\"transcription\"] = df[\"transcription\"].str.strip().str.strip('/-0123456789')\n",
    "    \n",
    "    # Ensure all commas and periods are followed by a space\n",
    "    df[\"transcription\"] = df[\"transcription\"].str.replace(\".\", \". \").str.replace(\",\", \", \")\n",
    "    \n",
    "    # Remove any instances of multiple spaces\n",
    "    df[\"transcription\"] = df[\"transcription\"].str.split().str.join(\" \")\n",
    "    \n",
    "    \n",
    "    # Add metrics\n",
    "    \n",
    "    # Grade level\n",
    "    df[\"grade_level\"] = df[\"story_id\"].str[0].astype(int)\n",
    "\n",
    "    # Length of story\n",
    "    df[\"story_length\"] = df[\"transcription\"].str.len()\n",
    "    \n",
    "    # Average word length\n",
    "    word_count = (df[\"transcription\"].str.split()).str.len()\n",
    "    df[\"avg_word_len\"] = df[\"story_length\"] / word_count\n",
    "    \n",
    "    # Number of quotation marks\n",
    "    # quote_count = df[\"transcription\"].str.count('\"')\n",
    "    # df[\"quotes_ratio\"] = quote_count / df[\"story_length\"]\n",
    "    df[\"quotes_num\"] = df[\"transcription\"].str.count('\"') \n",
    "    \n",
    "    # Number of unique words, over 2 characters    \n",
    "    def over_two_chars(transcription):\n",
    "        \"\"\"Returns number of unique 2+ char words in transcription.\"\"\"\n",
    "        word_list = transcription.split()\n",
    "        word_set = set()\n",
    "        for x in word_list:\n",
    "            if len(x) > 2: \n",
    "                word_set.add(x)\n",
    "        return len(word_set)\n",
    "    \n",
    "    # unique_words_count = df[\"transcription\"].apply(over_two_chars)\n",
    "    # df[\"unique_word_ratio\"] = unique_words_count / df[\"story_length\"]\n",
    "    df[\"unique_words_num\"] = df[\"transcription\"].apply(over_two_chars)\n",
    "    \n",
    "    # Number of adjectives\n",
    "    def num_adj(transcription):\n",
    "        \"\"\"Returns number of adjectives in transcription.\"\"\"\n",
    "        tokens = nltk.word_tokenize(transcription)\n",
    "        pos = nltk.pos_tag(tokens)\n",
    "        adj_count = 1\n",
    "        for word in pos:\n",
    "            if word[1] == 'JJ':\n",
    "                adj_count += 1\n",
    "        return adj_count\n",
    "    \n",
    "    # adj_count = df[\"transcription\"].apply(num_adj)\n",
    "    # df[\"adj_ratio\"] = adj_count / df[\"story_length\"]\n",
    "    df[\"adj_num\"] = df[\"transcription\"].apply(num_adj)\n",
    "    \n",
    "    # Percentage of complex words, based on textstat's automated function\n",
    "    # Compares to list of 3000 most commonuly used English words\n",
    "    # Number of words in story that are not on that list are considered \"complex\"\n",
    "    # Note: prone to errors, may need to be iterated on to be production-grade\n",
    "    num_complex_words = df[\"transcription\"].apply(textstat.difficult_words)\n",
    "    df[\"percent_complex_words\"] = num_complex_words / word_count * 100\n",
    "    \n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate both API and human transcription metrics\n",
    "human_metrics = calc_metrics(human_t)\n",
    "api_metrics = calc_metrics(api_t)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(167, 9)\n"
     ]
    },
    {
     "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>story_id</th>\n",
       "      <th>transcription</th>\n",
       "      <th>grade_level</th>\n",
       "      <th>story_length</th>\n",
       "      <th>avg_word_len</th>\n",
       "      <th>quotes_num</th>\n",
       "      <th>unique_words_num</th>\n",
       "      <th>adj_num</th>\n",
       "      <th>percent_complex_words</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3132</td>\n",
       "      <td>Page 1 Once there was a little cheatah and the...</td>\n",
       "      <td>3</td>\n",
       "      <td>1378</td>\n",
       "      <td>5.122677</td>\n",
       "      <td>14</td>\n",
       "      <td>136</td>\n",
       "      <td>14</td>\n",
       "      <td>4.460967</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  story_id                                      transcription  grade_level  \\\n",
       "0     3132  Page 1 Once there was a little cheatah and the...            3   \n",
       "\n",
       "   story_length  avg_word_len  quotes_num  unique_words_num  adj_num  \\\n",
       "0          1378      5.122677          14               136       14   \n",
       "\n",
       "   percent_complex_words  \n",
       "0               4.460967  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(human_metrics.shape)\n",
    "human_metrics.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(167, 9)\n"
     ]
    },
    {
     "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>story_id</th>\n",
       "      <th>transcription</th>\n",
       "      <th>grade_level</th>\n",
       "      <th>story_length</th>\n",
       "      <th>avg_word_len</th>\n",
       "      <th>quotes_num</th>\n",
       "      <th>unique_words_num</th>\n",
       "      <th>adj_num</th>\n",
       "      <th>percent_complex_words</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3132</td>\n",
       "      <td>Page. I 3132 Once there was a little cheatah a...</td>\n",
       "      <td>3</td>\n",
       "      <td>1375</td>\n",
       "      <td>5.092593</td>\n",
       "      <td>6</td>\n",
       "      <td>138</td>\n",
       "      <td>15</td>\n",
       "      <td>4.814815</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  story_id                                      transcription  grade_level  \\\n",
       "0     3132  Page. I 3132 Once there was a little cheatah a...            3   \n",
       "\n",
       "   story_length  avg_word_len  quotes_num  unique_words_num  adj_num  \\\n",
       "0          1375      5.092593           6               138       15   \n",
       "\n",
       "   percent_complex_words  \n",
       "0               4.814815  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(api_metrics.shape)\n",
    "api_metrics.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['friendo',\n",
       " 'dylanto',\n",
       " 'familys',\n",
       " \"lion's\",\n",
       " 'usall',\n",
       " 'parents',\n",
       " 'thier',\n",
       " 'hunting',\n",
       " 'dylans',\n",
       " 'bacle',\n",
       " 'dylan',\n",
       " 'minutes',\n",
       " 'bylans']"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Look at output of complex words feature -- note how many are not real words\n",
    "sample_transcription = api_metrics.iloc[0,1]\n",
    "textstat.difficult_words_list(sample_transcription)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'Page. I 3132 Once there was a little cheatah and the cheatah had a best friend lion the cheatah\\'s name was paws and the lion\\'s name was Dylan they always played with each other after they went hunting Dylan went to play with Paws after hunting with pack, and Paws wont to play after hunting with his mom. They always met up near the same rock arth the same lake, they would talk and have water fights here is what the talked about your parents and my pack might have a fight\" said Dylan \"We both don\\'t want that to happen we might have to fight against each other, said Paws After afew minutes it was time for them bacle to their family. Next morning they saw there family having a fight, Dylans family, won and Paus mom was hert. At thier usall he meeting time they talked aboutit. \"My mom got hert, It\\'s not fair\" said Paws I think the best way to fix this is to tell are family we are friendo\" said Dylan \"I think your right said Paws. So after there talk time and water fight they went back to there family and told We have been friends since 3 years ago said Paws and Dylanto their familys. Okay thier parents said I will let you goy swich spot for one day and we will see. to them. Pagea 3132 They were going to fet Paws spend time with Bylans pack and Dylan spend time with Paws mom it went well and everyone was happy and now the familys all went to the lake, The End'"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Compare to full transcription\n",
    "# Decision was made to keep compplex_words feature out of Squad Score for \n",
    "# unreliability in the limitations of this product\n",
    "sample_transcription"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Change this variable to alter which features are included in scaler and Squad Score\n",
    "scaled_features = [\"story_length\", \"avg_word_len\", \"quotes_num\", \"unique_words_num\", \"adj_num\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MinMaxScaler()"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scaler_h = MinMaxScaler()\n",
    "scaler_h.fit(human_metrics[scaled_features])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['../project/app/utils/complexity/MinMaxScaler.pkl']"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scaler_api = MinMaxScaler()\n",
    "scaler_api.fit(api_metrics[scaled_features])\n",
    "\n",
    "# Uncomment line below to pickle scaler for future use\n",
    "joblib.dump(scaler_api, \"../project/app/utils/complexity/MinMaxScaler.pkl\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def scale_row(row, transcription_source):\n",
    "    \"\"\"\n",
    "    Takes one row and scales it according to corresponding scaler.\n",
    "    \n",
    "    This function is integrated into the squad_score function\n",
    "    \n",
    "    row: full row of dataframe or a list\n",
    "    transcription_source (str): \"api\" or \"human\" based on source of transcription\n",
    "    \"\"\"\n",
    "    if transcription_source == \"api\":\n",
    "        scaler = scaler_api\n",
    "    elif transcription_source == \"human\":\n",
    "        scaler = scaler_h\n",
    "    else:\n",
    "        raise ValueError(\"Not a valid transcription source. Valid options: 'api' or 'human'.\")\n",
    "    \n",
    "    return scaler.transform([row[scaled_features]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.36490589, 0.45999584, 0.13953488, 0.34146341, 0.28888889]])"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Test on API transcription sample\n",
    "test_row = api_metrics.loc[0,scaled_features]\n",
    "scale_row(test_row, \"api\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.34876246, 0.47520045, 0.16666667, 0.35472973, 0.31707317]])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Test on human transcription sample\n",
    "test_row = human_metrics.loc[0,scaled_features]\n",
    "scale_row(test_row, \"human\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "story_length           1378\n",
       "avg_word_len        5.12268\n",
       "quotes_num               14\n",
       "unique_words_num        136\n",
       "adj_num                  14\n",
       "Name: 0, dtype: object"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check test_row\n",
    "test_row"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "def squad_score(row, transcription_source):\n",
    "    \"\"\"\n",
    "    Scales, weights, and adds all metrics for a given transcription.\n",
    "    \n",
    "    Scales according to transcription source (see scale_row function).\n",
    "    Weights according to presribed weights used only for Story Squad analysis.\n",
    "    \n",
    "    row: full row of dataframe or a list\n",
    "    transcription_source (str): \"api\" or \"human\" based on source of transcription\n",
    "    \"\"\"\n",
    "    \n",
    "    # Instantiate weights\n",
    "    weights = {\"grade_level\": 0,\n",
    "              \"story_length\": 1,\n",
    "              \"avg_word_len\": 1,\n",
    "              \"quotes_number\": 1,\n",
    "               \"unique_words\": 1,\n",
    "               \"adj_num\": 1,\n",
    "              \"percent_complex_words\": 0}\n",
    "    \n",
    "    # Scale needed metrics\n",
    "    scaled = scale_row(row, transcription_source)[0]\n",
    "    \n",
    "    # Generate scaler to create desired output range (~0-100)\n",
    "    range_scaler = 30\n",
    "    \n",
    "    # Weight values\n",
    "    # gl = weights[\"grade_level\"] * row[2] * range_scaler\n",
    "    sl = weights[\"story_length\"] * scaled[0] * range_scaler\n",
    "    awl = weights[\"avg_word_len\"] * scaled[1] * range_scaler\n",
    "    qn = weights[\"quotes_number\"] * scaled[2] * range_scaler\n",
    "    uw = weights[\"unique_words\"] * scaled[3] * range_scaler\n",
    "    an = weights[\"adj_num\"] * scaled[4] * range_scaler\n",
    "    # pcw = weights[\"percent_complex_words\"] * scaled[5] * range_scaler\n",
    "    \n",
    "    # Add all values\n",
    "    s_score = sl + awl + qn + uw + an\n",
    "    \n",
    "    return s_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate Squad Scores for all transcriptions and generate new DFs\n",
    "h_squad_score = human_metrics[[\"story_id\"]].copy()\n",
    "h_squad_score[\"squad_score\"] = human_metrics.apply(lambda x: squad_score(x, \"human\"), axis=1)\n",
    "\n",
    "api_squad_score = api_metrics[[\"story_id\"]].copy()\n",
    "api_squad_score[\"squad_score\"] = api_metrics.apply(lambda x: squad_score(x, \"api\"), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>story_id</th>\n",
       "      <th>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3132</td>\n",
       "      <td>47.843668</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  story_id  squad_score\n",
       "0     3132    47.843668"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Check head\n",
    "api_squad_score.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>167.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>50.095676</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>20.209209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.731707</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>35.305735</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>47.974608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>60.715916</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>124.558085</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       squad_score\n",
       "count   167.000000\n",
       "mean     50.095676\n",
       "std      20.209209\n",
       "min       0.731707\n",
       "25%      35.305735\n",
       "50%      47.974608\n",
       "75%      60.715916\n",
       "max     124.558085"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "h_squad_score.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "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>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>167.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>47.399637</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>19.397492</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.668999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>33.216693</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>43.596373</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>58.944506</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>119.110260</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       squad_score\n",
       "count   167.000000\n",
       "mean     47.399637\n",
       "std      19.397492\n",
       "min       1.668999\n",
       "25%      33.216693\n",
       "50%      43.596373\n",
       "75%      58.944506\n",
       "max     119.110260"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "api_squad_score.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize distribution of Squad Score for human transcriptions\n",
    "ax = sns.distplot(h_squad_score[\"squad_score\"])\n",
    "plt.title(\"Distribution of Squad Scores in Human Transcriptions\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize distribution of Squad Score for API transcriptions\n",
    "# Note that API distribution is similar to human distribution\n",
    "ax = sns.distplot(api_squad_score[\"squad_score\"])\n",
    "plt.title(\"Distribution of Squad Scores in API Transcriptions\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEXCAYAAAC9A7+nAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3hc5ZX48e8ZdVnNsprVLHdbNm4YG1NNMTGhpRACBAIsCyEbUjbJ7ia7CSGkbLLZsNkNpBCSXxICcYCEYMCmdwJuuEq2bLlKsprVJavP+f0xIzIIyRrZM7ozo/N5Hj2ecue+xyPp6M657z2vqCrGGGPCn8vpAIwxxgSGJXRjjIkQltCNMSZCWEI3xpgIYQndGGMihCV0Y4yJEJbQxzkR+YWIfDNA+yoUkXYRifLef1VE/jEQ+/bub72I3BSo/Y1i3O+KyDERqRnrsb3jHxKRi0/h9UUioiISHci4hhjnlOI0py6o32DjLBE5BGQDfUA/UAr8HnhAVd0AqnrHKPb1j6r64nDbqOoRIOnUon5vvLuBGap6g8/+Lw3EvkcZRyHwFWCKqtaN9fjGjIYdoUe+K1Q1GZgC/AD4N+DXgR4k2Ed/DioEGgKVzAc+vRgTDJbQxwlVbVHVtcAngZtEZD6AiPxWRL7rvZ0hIk+LSLOINIrIGyLiEpGH8CS2p7wllX/1+Rh/q4gcAV4e5qP9dBHZKCKtIvKkiKR7x1opIpW+MQ58ZBeR1cC/A5/0jrfd+/x7JRxvXN8QkcMiUicivxeRVO9zA3HcJCJHvOWS/xjuvRGRVO/r6737+4Z3/xcDLwC53jh+O8zr/1VEqkXkqIj8o3fsGT7v789FZJ2IdAAXiMhlIrLV+55UeD+N+O7vRm8cDYPj9sb1NRHZ733+0YH31F/e/++vvTFXeUtKUSIS5/3ez/fZNlNEOkUky3v/chHZ5t3ubyKyYDRjm+CyhD7OqOpGoBI4d4inv+J9LhNPqebfPS/RG4EjeI72k1T1v3xecz4wF/jQMEN+GvgHYDKe0s//+RHjs8D3gT95x1s4xGY3e78uAKbhKfXcN2ibc4DZwEXAXSIyd5ghfwqkevdzvjfmW7zlpUuBo944bh78Qu8fny8DFwMzgJVD7P964HtAMvAm0OEdIw24DPisiHzEu79i4OfAjUAuMAnI99nX54GPeOPMBZqA+4f5fw3nt3i+FzOAxcAleMpp3cBfgOt8tr0GeE1V60RkMfAb4DPeuH4JrBWRuFGOb4LEEvr4dBQY6qiuF0/inaKqvar6ho7c7OduVe1Q1c5hnn9IVXepagfwTeCaAJUdPgXcq6oHVLUd+Dpw7aBPB99W1U5V3Q5sBz7wh8Eby7XA11W1TVUPAT/Gk1D9cQ3w/1S1RFWPA3cPsc2TqvqWqrpVtUtVX1XVnd77O4A/4knQAFcDT6vq694E+03A7bOvO4D/UNVK7/N3A1f7W/ISkWzgw8CXvN+3OuB/vO8BwCM+t8Hzx+gR7+3bgV+q6gZV7VfV3wHdwJn+jG2CzxL6+JQHNA7x+I+AcuB5ETkgIl/zY18Vo3j+MBADZPgV5Ynlevfnu+9oPJ8sBvjOSjnO0CdsM7wxDd5X3iji8P0/DvV+vO8xEVkuIq94SzwteJL0wHvyvv15/xA2+Lx8CvCEt+TRDOzGc8Lb9/99IlPw/H+rffbxSyDL+/wrQKI3xiJgEfCEz2u/MvA672sLvDGbEGAJfZwRkTPwJKs3Bz/nPUL9iqpOA64EviwiFw08PcwuRzqCL/C5XYjnU8AxPGWHRJ+4ovCUevzd71E8CcZ3331A7QivG+yYN6bB+6ry8/XVvL8kUjDENoP/L48Aa4ECVU0FfgGIz/7e24eIJOIpbwyoAC5V1TSfr3hV9TfeCjxH1Rk+r09R1XkAqtoPPIqn7HIdnk8LbT6v/d6gsRNV9Y9+jm2CzBL6OCEiKSJyObAG+IOq7hxim8tFZIaICNCC58hv4ON+LZ4a82jdICLF3sR0D/C4N2nsBeK9JwhjgG8AvrXYWqBIRIb7Gf0j8M8iMlVEkvh7zb1vNMH5JLDviUiyiEzBUxP/g5+7eBS4RUTmev+P/szpTwYaVbVLRJbhKWsMeBy4XETOEZFYPO+Z73vwC2+sU+C9k5ZX+RkrqloNPA/82Psz4RKR6SJyvs9mj+A5ef4p/l5uAfgVcIf36F1EZIL3+5fs7/gmuCyhR76nRKQNz9HVfwD3ArcMs+1M4EWgHXgb+JmqvuJ97j+Bb3g/an91FOM/hOckXA0QD3wBPLNugH8CHsRzNNyB54TsgMe8/zaIyLtD7Pc33n2/DhwEuvCcMDwZn/eOfwDPJ5dHvPsfkaqux3Oi9xU85ap3vE91n+Bl/wTc4/2+3IXnj8LA/kqAz3ljqMZz0tP3fflfPEf3z3tf/w6w3J9YfXwaiMVzXUITnj8ik31i2IDn/cgF1vs8vhm4Dc/J5ybv//fmUY5tgkhsgQtjAsc7k2YXEDfaTwvGnCo7QjfmFInIR71zuCcCPwSesmRunGAJ3ZhT9xmgDtiP57zDZ8c6ABH5lPfip8FfJWMdi3GOlVyMMSZC2BG6McZECMcaKmVkZGhRUZFTwxtjTFjasmXLMVXNHOo5xxJ6UVERmzdvdmp4Y4wJSyJyeLjnrORijDERwhK6McZECEvoxhgTISyhG2NMhLCEbowxEcKvhC4iq0WkTETKh+qRLZ7V3l/xLqu1Q0Q+HPhQjTHGnMiICd3bp/p+PEtxFQPXeZfJ8vUN4FFVXYxntZOfBTpQY4wxJ+bPEfoyoNy71FcPnn7ag/svK5DivZ2KZ/EBY4wxY8ifC4vyeP8SWpV8sP/y3Xj6M38emIBnwVxjjDFjKFBXil4H/FZVfywiK4CHRGS+qvoubouI3I5noVkKCwsDNLQx4eeRDUecDuGUXb/cfodDjT8llyrev05iPh9cb/FWvKuuqOrbeFam+cBCwKr6gKouVdWlmZlDtiIwxhhzkvxJ6JuAmd61G2PxnPRcO2ibI8BF8N6KLfFAfSADNcYYc2IjJnTvyit3As8Bu/HMZikRkXtE5ErvZl8BbhOR7XgW771ZrdG6McaMKb9q6Kq6Dlg36LG7fG6XAmcHNjRjjDGjYVeKGmNMhLCEbowxEcISujHGRAhL6MYYEyEsoRtjTISwhG6MMRHCEroxxkQIS+jGGBMhLKEbY0yEsIRujDERwhK6McZECEvoxhgTISyhG2NMhLCEbowxEcISujHGRAhL6MYYEyEsoRtjTITwK6GLyGoRKRORchH52hDP/4+IbPN+7RWR5sCHaowx5kRGXIJORKKA+4FVQCWwSUTWepedA0BV/9ln+88Di4MQqzHGmBPw5wh9GVCuqgdUtQdYA1x1gu2vw7NQtDHGmDHkzyLReUCFz/1KYPlQG4rIFGAq8PKph2aC4ZENR5wO4ZRdv7zQ6RCMCUmBPil6LfC4qvYP9aSI3C4im0Vkc319fYCHNsaY8c2fhF4FFPjcz/c+NpRrOUG5RVUfUNWlqro0MzPT/yiNMcaMyJ+EvgmYKSJTRSQWT9JeO3gjEZkDTATeDmyIxhhj/DFiQlfVPuBO4DlgN/CoqpaIyD0icqXPptcCa1RVgxOqMcaYE/HnpCiqug5YN+ixuwbdvztwYRljjBktu1LUGGMihCV0Y4yJEJbQjTEmQlhCN8aYCGEJ3RhjIoQldGOMiRCW0I0xJkJYQjfGmAhhCd0YYyKEJXRjjIkQltCNMSZCWEI3xpgIYQndGGMihCV0Y4yJEJbQjTEmQlhCN8aYCGEJ3RhjIoRfCV1EVotImYiUi8jXhtnmGhEpFZESEXkksGEaY4wZyYhL0IlIFHA/sAqoBDaJyFpVLfXZZibwdeBsVW0SkaxgBWyMMWZo/hyhLwPKVfWAqvYAa4CrBm1zG3C/qjYBqGpdYMM0xhgzEn8Seh5Q4XO/0vuYr1nALBF5S0TeEZHVQ+1IRG4Xkc0isrm+vv7kIjbGGDOkQJ0UjQZmAiuB64BfiUja4I1U9QFVXaqqSzMzMwM0tDHGGPCjhg5UAQU+9/O9j/mqBDaoai9wUET24knwmwISpTHGcarK7uo2Ximro9+tPFdSQ3ZKHFefXsCyqelOh2fw7wh9EzBTRKaKSCxwLbB20DZ/xXN0johk4CnBHAhgnMYYB1W3dPLrtw7yhw2H6elzM3FCLM3He3i+tJZrH3ibX7y2H1V1Osxxb8QjdFXtE5E7geeAKOA3qloiIvcAm1V1rfe5S0SkFOgH/kVVG4IZuDFmbOyvb+e3bx0iNtrFFQtzWVaUTpRLuH55Ie3dffzbn3fwg/V72HK4if/+xEJSE2KcDnnc8qfkgqquA9YNeuwun9sKfNn7ZYyJEI0dPfxx4xHSk2L5zLnTSIx7f8pIiovmvusWc3rhRL6/bje3/X4zf7ztTKJc4lDE45tdKWqMGVJ3bz8PvXMItyo3njnlA8l8gIjwD+dM5T8/dhobDzbyy9f3j3GkZoAldGPMB7hVeWxLJXWt3Vy3rJCMpLgRX3P16flcdtpk7n1+Lzsqm8cgSjOYJXRjzAfsqGymtLqVS+fnMDMr2a/XiAjf++h8MpPj+NKabRzv6QtylGYwS+jGmPfp7XfzfGktuanxnDUjY1SvTUuM5cfXLORgQwc/XL8nSBGa4VhCN8a8zzsHGmg+3svq+ZNxyehPbp41PYPrlxXy8IYjVDYdD0KEZjiW0I0x7zne08crZXXMyk5iRlbSSe/nzgtn4BLh/lfKAxidGYkldGPMe14tq6e7183qeZNPaT+TUxO4dlkBj22upKLRjtLHiiV0YwwAzcd7ePtAA0sKJ5KTGn/K+/vsyum4RPjZq3aUPlYsoRtjAHh7fwOqykVzA7OcweTUBK6zo/QxZQndGENPn5vNh5sonpxCWmJswPb72ZUzcLmslj5WLKEbY9he2Uxnbz8rpo9umuJIclLjufr0fJ7YWkXL8d6A7tt8kCV0Y8Y5VeXt/Q3kpMRTNCkx4Pu/flkh3X1unthaGfB9m/ezhG7MOHeo4Tg1rV2smDYJOYl55yOZn5fKwvxUHtl4xFrsBpkldGPGubf3HyMhJoqFBR9YZCxgrltWyN7adt490hS0MYwldGPGtZbOXkqrW1k6ZSKx0cFLB1cszCUpLpqHNxwJ2hjGz37oJny53crDG4+wfmc1x9q7qWzqJDbKxTkzMzijKJ2YKPubPp5tPdKEW2H5tElBHWdCXDRXLcrl8S2VfOvyeaQm2iIYwWC/zRGsovE4N/x6A9/86y4aO3qYmjGB+XmppCXG8vSOav77+TLeOdBgdc1xSlXZWtHMlEmJpE8I3FTF4Vy/3HNy9C92cjRo/DpCF5HVwP/iWYLuQVX9waDnbwZ+xN8Xj75PVR8MYJxmlNbtrOZfHtsOwH9+7DSuPaMAEeER70feA/XtvLi7jrXbj9Le3cfFc7OdDNc4oLqli/q2bq5alDsm483L9ZwcXbOxglvOnjomY443Ix6hi0gUcD9wKVAMXCcixUNs+idVXeT9smTuoC2Hm/jSmm3Mzknm2S+dx3XLCj8we2FaZhK3nTuV0wsn8vKeOt4qP+ZQtMYp2yqaiRLhtLzUMRvz46fnU1bbRllN25iNOZ74U3JZBpSr6gFV7QHWAFcFNyxzsmpaurjjD1vISY3nNzefQUH68POKRYSPLM5jXm4Kz+ysZsvhxjGM1DjJrcr2ymZm5SSTGDt2p9IunT8Zl8DTO46O2ZjjiT8JPQ+o8Llf6X1ssI+LyA4ReVxECobakYjcLiKbRWRzfX39SYRrTqSrt5/P/GELHd19/OrTS/26hDvKJXxyaQEzspL4y7tV1r96nDhQ30FbVx+LgjhVcSiZyXGcNT2Dp7YftXM3QRCok6JPAUWqugB4AfjdUBup6gOqulRVl2ZmZgZoaDPgO0+Xsr2imXuvWcTsHP+WDQOIjnJx/bJCkuOjeWJrFf1u+0WLdNsqmoiLdjFnFD8ngXLFwskcajjOrqrWMR870vmT0KsA3yPufP5+8hMAVW1Q1W7v3QeB0wMTnvHXjspmHt5whFvPmcrq+Tmjfn18TBSXL8iluqWLt/dbPT2S9fS5KTnayvy8VEemrX5oXg4xUcJTVnYJOH++m5uAmSIyVURigWuBtb4biIhvN/wrgd2BC9GMRFW556lSMpJi+dLFM096P/NyU5iTk8wLu2tpOt4TwAhNKNlT00p3n3vMyy0D0hJjOW9mJk9vP4rbPg0G1IgJXVX7gDuB5/Ak6kdVtURE7hGRK72bfUFESkRkO/AF4OZgBWw+6Okd1Ww+3MRXL5lNcvzJX7AhIlyx0DOFzWqckavkaCtJcdFMzZjgWAxXLMzlaEuXtQIIML9Ob6vqOmDdoMfu8rn9deDrgQ3N+KOrt58frN/D3MkpfGLpkOeiR2ViYiwXz81m/a4ayuvamZk99jVWEzy9/W7KatpYWJB2UgtAB8rFxdnERbt4avtRlhalOxZHpLErRcPcr14/QFVzJ9+6opgoV2B+QVdMn0RqQgwvl9XZUXqEKa9rp6ffzbzcFEfjSIqL5qK5WTyzs8ZOwgeQJfQw1tbVywOvH+CS4mzODGAvjmiXi/NmZnC44TgHGzoCtl/jvNKjrcTHuJiW6Vy5ZcDq+ZM51t7NtgoruwSKJfQw9seNR2jr7uPOC2cEfN9Li9JJiovm1T12vUCk6HcrpdWtzMlJIdrl/K/+ytmZxEQJz5fUOh1KxHD+u2pOSk+fm9+8eYgzp6WzID/wsxViolycOzOD8vp2jtgCvxHhUEMHnb39FE92ttwyICU+hjOnTeL50lor7QWIJfQw9dT2o9S0dvGZ86YHbYxlU9NJiIni1bK6oI1hxk7J0VZiooRZIXSi+5J5ORw81sH++nanQ4kIltDDkKryqzcOMCs7iZWzg3fFbVx0FGfPyGBPTRvVLZ1BG8cEn1uV0qMtzMxKDupCFqO1ytvl8/lSK7sEQuh8Z43fXttbz56aNm47d1pQ1oD0tWLaJGKihHcOWOOucFbV1ElrV5/js1sGy0mNZ2F+qtXRA8QSehj61RsHyE6J46pFQ/VIC6yE2CgW5KWxvaKZrt7+oI9ngqO0uhWXwJyc0EroAKuKs9lW0Uxta5fToYQ9S+hh5kB9O2+VN/DpFUVj9tF52dR0evrdbKtoHpPxTODtqWmlaNIEEmKjnA7lAy6Z5+k99OJuO0o/VZbQw8yfNlcQ5RI+cXr+mI2ZPzGB3LR4Nh5stNkIYaipo4fa1m5HOiv6Y2ZWEkWTEq3sEgCW0MNIb7+bP2+p5MI5WWSlxI/ZuCLC8qJJ1LR22RTGMLSn1rM6UCiWW8Dz87WqOJu39zfQ3t3ndDhhzRJ6GHlpdy3H2nu49oxT79kyWgsKUomLdrHhoJ0cDTdlNa1MmhBLRnKc06EM68I52fT0u20pxFNkCT2MrNlUQU5KPOfPGvvFQeKio1hcmMauqhY67CgqbHT39bO/viNkyy0DlhZNJDk+mlf22DUPp8ISepg42tzJa3vr+cTSfKIdWJQAYNnUSfS51U6OhpH9dR30u5U5IXJ16HBiolycNzOTV6wh3CmxhB4mHt3sWdb1mgC0yD1ZOSnx5KUlsNWaKYWNPTWtxEW7mDJp+MXCQ8UFc7Kobe2m5KgtTXeyLKGHAbdbeWxzJefMyKAg3dlfzEUFaRxt7rI5w2HArUpZbRszs5NDohnXSAauerayy8kL/e+yYeOhRqqaO7l6DKcqDsezMAJWdgkD1c1dtHX1hXz9fEBGUhwL81N52XoHnTS/ErqIrBaRMhEpF5GvnWC7j4uIisjSwIVontxWRWJsFKuKs50OhaS4aGZmJbOtohm31TpD2p6aVgRCqhnXSC6Yk8W2imYa2rtH3th8wIgJXUSigPuBS4Fi4DoRKR5iu2Tgi8CGQAc5nnX39fPMjmo+NC+HxFi/VgwMukWFabR09nLwmC1+EcrKatvIn5hAUlxo/Nz448I5Wah6+hWZ0fPnCH0ZUK6qB1S1B1gDXDXEdt8BfghYcTWAXi2rp7WrjysX5TodynuKJ6cQF+1i6xEru4Sq9u4+qpo6mRUm5ZYB83NTyUiK42Wro58UfxJ6HlDhc7/S+9h7RGQJUKCqz5xoRyJyu4hsFpHN9fX2F9gfa7cdZdKEWM6dkeF0KO+JiXIxPy+VXUdb6OlzOx2OGUJ5XRsKzA6jcguAyyVcMDuT1/fW09dvP1ujdconRUXEBdwLfGWkbVX1AVVdqqpLMzPH/uKYcNPW1cuLu2u5fMFkx+aeD2dxQRo9fW5Kq22KWSgqq2ljQmwUuWkJTocyahfMyaK1q4+tduJ91PzJElWA7+TnfO9jA5KB+cCrInIIOBNYaydGT92zu2ro7nNz1eLgt8kdraKMCaTER7Oz0n7pQo1blX117czKTsYV5H75wXD2jAyiXGIrZZ0EfxL6JmCmiEwVkVjgWmDtwJOq2qKqGapapKpFwDvAlaq6OSgRjyNPbjvKlEmJLC4I/Jqhp8olwml5qeyta6ezx/qkh5Kqpk6O9/SH1ewWX6kJMZxeOJFXy6wsO1ojJnRV7QPuBJ4DdgOPqmqJiNwjIlcGO8Dxqq6ti7/tP8aVC3ODvirRyTotP41+t7Lbyi4hpay2DcHTljZcnT87k5KjrdS12RyL0fCrMKuq61R1lqpOV9XveR+7S1XXDrHtSjs6P3XP7arBrXDFwtCZ3TJYwcQE0hJi2FnV4nQoxsfe2jYK0hNJDKPpioMNXDX6mh2lj0ponWkz73lmZzUzspJC+mOzeMsu++raON5jHRhDwXvTFbPD9+gcPFNjs5LjeNXmo4+KJfQQVNfWxYaDjVx22mSnQxnRafmpuBVKraFSSNhX65muGMoHAv4QEc6flckbNn1xVCyhh6Bnd9WgCpctCP2EnpeWQPqEWHZY2SUk7K1tY0JcdFhOVxxs5WzP9EXrG+Q/S+gh6Jkd1cwM8XLLgIGyy4H6dls+zGHvTVfMSgrL6YqDnTNzYPqilV38ZQk9xNS1drHxUCMfDoNyy4DT8jxll5KjdpTupHCfrjhYakIMSwrTeHWvzUf3lyX0EPNsSfiUWwZMTo1n0oRYSqqsju6kSJiuONjK2VnsqrLpi/6yhB5ing6jcssAEWF+XioHjrVz3Moujtnr7a4YztMVBxtYP/f1vbZ4tD8soYeQurYuNoVZuWXA/FzvbBe7yMgR4dpdcSTzclPITI6zNgB+soQeQp4rqQ27csuA3LR4JibGsMvq6I4or2sPy+6KI3lv+uK+YzZ90Q+W0EPI+p3VTMucEJY1UBFhfm4q++s6rLeLA/bWtpEYpt0VR7JydiYtnb1st0ZwI7KEHiIaO3rYcLCRS+fnhGzvlpHMy0ulX5U9NVZ2GUtuVfbVtoVtd8WRnDsjE5dg0xf9YAk9RLxQWkO/W7l0fviVWwbkT0wgNSGGXXaR0Zg62txJR09/2F/uP5zUxBiWWPdFv1hCDxHrd9VQkJ7AvNwUp0M5aS4R5uWmsK+une5eK7uMlb9PV4ys+rmvlbMz2VnVQn2bLR59IpbQQ0BLZy9vlR/j0vmTw7bcMmBebip9bmVPbZvToYwbe2vayJuYwIQImq442MrZWQC8bs26TsgSegh4aXctvf3K6vk5TodyyqZMSiQ5LpoSK7uMiY7uPiqbOiNudstgxZNTyEiy7osjsYQeAtbvqiEnJZ5F+aG3MtFouUQozk2hrLbNFpAeA/sGFoOOsPnng7lcA9MX6+l3q9PhhCxL6A7r6O7j9b31rJ6fg8sV3uWWAfNyU+ntV/Za2SXo9tRETnfFkaycnUnz8V62VTQ5HUrI8iuhi8hqESkTkXIR+doQz98hIjtFZJuIvCkixYEPNTK9UlZHd587IsotA6ZmTCAxNsqadQWZZ7piO7OzI6O74kjOm5lJlEt4eY9dNTqcERO6iEQB9wOXAsXAdUMk7EdU9TRVXQT8F3BvwCONUOt31ZCRFMsZRelOhxIwUS5h7uQU9tS02dV9QVTReJzO3sjprjiS1MQYTp8ykZf3WB19OP4coS8DylX1gKr2AGuAq3w3UFXfK0kmAFbk8kNXbz+v7Knjknk5REVIuWXA/NwUuvvclNe3Ox1KxCqrbcMlkT1dcbAL52Sxu7qV6pZOp0MJSf4k9Dygwud+pfex9xGRz4nIfjxH6F8YakcicruIbBaRzfX19lf2tb31HO/p58NhfDHRcKZnJhEX7bKWukFUVtNGYfoEEmKjnA5lzFw4xzN98RU7Sh9SwE6Kqur9qjod+DfgG8Ns84CqLlXVpZmZmYEaOmyt31lNWmIMy6dFTrllQHSUi7mTUyitbrVZCUHQ0tlLdUtXxM9uGWxmVhJ5aQlWRx+GPwm9CijwuZ/vfWw4a4CPnEpQ40F3Xz8v7a5j1dxsYqIic7LRvNwUOnv7OXisw+lQIs4+7wyiSJ9/PpiIcOGcLN4qP0aXXY38Af5kkk3ATBGZKiKxwLXAWt8NRGSmz93LgH2BCzEy/a28gbbuvrDsfe6vWdnJxEa5rLdLEOypaSM1IYbslDinQxlzF87JorO3nw0HG50OJeSMmNBVtQ+4E3gO2A08qqolInKPiFzp3exOESkRkW3Al4GbghZxhFi3s5rkuGjOmjHJ6VCCJibKxeycZEqqW3GrlV0CpbffTXldO7Ozk8O+VcTJWDF9EvExLl6xsssH+NX8QVXXAesGPXaXz+0vBjiuiNbb7+aF3bVcNDeLuOjIPqE1Py+VnVUtHDzWwfTMyOwGONYOHuugp9/N3Mnjq9wyID4mirOmZ/Dynjq+dUXxuPyjNpzILN6GuHcONNB8vJfVETi7ZbDZ2cnERImVXQJod3UrMVHCtHH8B/KCOVkcaTzO/no7P+PLEroD1u2sZkJsFCtnR/5Mn9hoF7Ozkyk5amWXQFBV9tS0MTMrOWJPpvvjIu/0xRd31zocSWgZvz8RDuntd/PsrhoumptNfExkl1sGzM9Lpb27j0MNdjR1qqpbumjp7B235ZYBuWmetQNeLLWE7ssS+hh7e38DTcd7w3Ih6JM1O8fKLoGyu7oVAWbnhO9CKIGyqjibLUeaONZui14MsIQ+xgbKLefPigz4PJUAABmWSURBVPxyy4C46ChmZSdTUmVll1O1u6aVgvREkiJ4MQt/rSrORhVe3m2zXQZYQh9Dvf1uni2p4eLi8VNuGTA/L5W27j4ONxx3OpSw1dLZy9HmLuaOs6tDh1M8OYW8tASet7LLeyyhj6G393tmt1wWwRcTDWdOTjLRLiu7nIo9NZ6+OHMmW7kFPFeNrirO5s3yejp77KpRsIQ+pp7ZUU1SXDTnjaNyy4C46Chm5ySzq6rFyi4naXd1K+kTYslKHn9Xhw5nVXE2Xb1u3thnzbrAEvqY6e1381xpDRfPzRp35ZYBC/LTaOvus94uJ6Grt5/99R3MzRmfV4cOZ9nUdFLio3nByi6AJfQx81b5MU+5ZUGu06E4Zk5OMrHRLnZUNjsdStjZU9NGv1uZn5fqdCghJSbKxQVzsnh5T5119cQS+phZu/0oyfHRnDcrw+lQHBMT5aJ4cgq7qlrpc9tKRqNRcrSF5PhoCtITnQ4l5Kwqzqaho4d3j9hao5bQx0BXbz/P7arhw/MnR3zvlpEsyE+ls7ef8lpbychfPX1u9ta2UTw5ZVysHTpaK2dnERvtYv3OGqdDcZwl9DHw0u46Onr6uWrR+C23DJiRlURCTBQ7bLaL3/bWttHbb+WW4STFRXPezEzW76rGPc7LLpbQx8CT26rISo5j+bTIbZXrr2iXi/l5KZQebaWnz8ou/th1tIXE2CiKJk1wOpSQddmCHKpbuthaMb7Pz1hCD7KW4728WlbPFQtzI24h6JO1ID+Nnn73e/OqzfD6+t2U1XjKLfbzM7yL5mYTG+Vi3c5qp0NxlCX0IFu/q5qefreVW3xMzZhAcnw0Oyqt7DKS8rp2uvvcVm4ZQUp8DOfNymD9zvFddrGEHmRPbjvK1IwJnGa/kO9xibAgL5WymjaOd/c5HU5I23W0lfgYF9Myrdwykg+fNpmjLV1sG8fTYv1K6CKyWkTKRKRcRL42xPNfFpFSEdkhIi+JyJTAhxp+alq6eOdgA1ctyrWLQQZZMmUi/apst5Ojw+pzu9ld3crcnBSiXXbsNZKLi71llx3jt+wy4k+JiEQB9wOXAsXAdSJSPGizrcBSVV0APA78V6ADDUdrt1ehClcutHLLYJNTE8hJiWerzR0e1r7adjp7+1mQb5/u/JESH8O5MzNYv6sGHaftJfz5s78MKFfVA6raA6wBrvLdQFVfUdWBNnrvAPmBDTP8qCqPba5kcWHauF4q7ESWFKZR2dRJXVuX06GEpG0VzSTGRjEjy7or+uvDp02mqrmTbeN0tos/CT0PqPC5X+l9bDi3AuuHekJEbheRzSKyub4+spvpbK9sYV9dO584vcDpUELWwoI0XAJbj4zPX74T6e7tZ09NK6flpdrsllEYKLus3X7U6VAcEdDCnIjcACwFfjTU86r6gKouVdWlmZmR3XHwsc0VxMe4uHzh+GuV66/k+BhmZiWz9UiTdWAcpLS6ld5+ZVFBmtOhhJXUhBgunJPFU9uP0tc//q5z8CehVwG+h5n53sfeR0QuBv4DuFJVx/WaUF29/azdfpTV83JIiY9xOpyQtmTKRFq7+jhgq7e/z7aKZiYmxlBovVtG7aNL8jjW3sMb+445HcqY8yehbwJmishUEYkFrgXW+m4gIouBX+JJ5uN+PajnS2tp6+rjE0ut3DKSOTnJxMe4rLGSj7auXvbXt7MwP81mR52EC2ZnkZYYw1+2fuC4M+KNmNBVtQ+4E3gO2A08qqolInKPiFzp3exHQBLwmIhsE5G1w+xuXHhscwV5aQmssEv9RxQT5WJBfholR1ts1RmvnVUtuNVzjsGMXmy0iysW5PJ8SQ1tXb1OhzOm/Kqhq+o6VZ2lqtNV9Xvex+5S1bXe2xeraraqLvJ+XXniPUauo82dvFl+jI+fno/LTmb5ZVlROr39ytYKO0oH2F7RzOTUeLJT4p0OJWx9dEke3X3ucdeB0a5WCLA/b6lEFa5eMu5nbvotNy2B/IkJbDzYOG7nDw+oa+uioqnTToaeosUFaUzNmMBftlY6HcqYsoQeQH39bh7ZeIRzZ2ZQOMlOZo3GsqJ06tq6OdxwfOSNI9jmQ024BEvop0hE+MiiPN450EhVc6fT4YwZS+gB9NKeOqpburjhTOt8MFoL8tOIi3ax8VCj06E4pq/fzbtHmpg7OYVkmx11yj662HO5zBPvjp+jdEvoAfSHdw6TmxrPRXOynA4l7MRGu1hcOJFdVS3jtmFXaXUrx3v6OaMo3elQIkLhpERWTJvEmk0V42a9UUvoAXKgvp039h3j+uWFREfZ23oylk1Np8+t43YK4+ZDTaQlxjAjy1pFBMqnziyksqmT1/dG9pXpAyzzBMgf3jlCTJTwyTMKnQ4lbOWkxFOYnsiGg43j7srRxo4eyuvbOX3KRFs3NIAuKc4hIymOhzccdjqUMWEJPQCO9/Tx2JYKLp0/mczkOKfDCWsrpk2ioaOHvTVtTocypjYfakSApVOs3BJIsdEuPnlGPi/vqRsXJ0ctoQfA2m1Haevq48YVdjL0VM3PSyU1IYY3ysfPZdv9bmXLkSZmZSeTmmAnQwPt2jMKUeBPG484HUrQWUI/RW638uCbB5k7OYWlUyY6HU7Yi3IJZ02fxMFjHVQ1Rf4RFcCuqhbauvpYPtWOzoOhID2RlbMyWbOpgt4Ib9hlCf0UvbSnjvK6du44f5r13QiQM4rSiYt28UZ55J/IUlXeLD9GRlIss3Ks73mw3HDmFOraunmhtNbpUILKEvop+sVr+8mfmMBlp1mb3ECJj4nijKJ0dlW10Hy8x+lwgupgQwdVzZ2cPSPDToYG0crZWeSlJfD/3jrodChBZQn9FGw61MiWw03cdu40m6oYYGdN9zQ2+9v+BocjCa639h0jMTaKxQVWrgumKJdw6zlT2XSoiS2HI/fiNctCp+AXr+5nYmIM11ib3IBLS4xlfl4qmw41crwnMi80OtbWzZ6aNpZPnURstP0qBtu1ywpIS4zhF68dcDqUoLGfopNUVtPGS3vquOmsIhJio5wOJyKtnJVFT5+bNyN0oYK39h/D5RLOnGYnQ8dCYmw0n15RxAultZTXRea0WEvoJ+nnr5aTEBPFTSuKnA4lYuWkxjM/L5W/7W+gPcLaAbR39/HukSYWFaRZ35YxdNOKKcTHuPhlhB6lW0I/CXtqWnly+1E+vWIKEyfEOh1ORLtobha9/W7eiLBLt1/fW09fv3LuzAynQxlXJiXFcc3SAv66rYqali6nwwk4S+gn4b+fKyMpLprPrpzudCgRLys5nkUFabxzsCFiVp+pbe3inQMNLC5MIyvZFrEYa7edO41+t/LgG5F3lO5XQheR1SJSJiLlIvK1IZ4/T0TeFZE+Ebk68GGGji2HG3lxdx13nD+dtEQ7Oh8LF87Jot+tvBohR+k/fXkfblUunJPtdCjjUkF6Ih9ZnMfv3znM0QhrBzBiQheRKOB+4FKgGLhORIoHbXYEuBl4JNABhhJV5YfPlpGRFMctZxc5Hc64MSkpjiWFE9l4sJHGjvCel17ReJw1GytYWpROupXrHPPPF88ChZ+8uNfpUALKnyP0ZUC5qh5Q1R5gDXCV7waqekhVdwARfV3ta3vr2XiwkS9eNIPE2GinwxlXLpqbjUtg3c5qp0M5Jf/70j5cLuGC2dYz30kF6YncuGIKj2+pZF9t5Mx48Seh5wEVPvcrvY+NmojcLiKbRWRzfX14fXzu7Xfzn+v2UJieaC1yHZCaEMOFs7MorW7l1bI6p8M5KXtr2/jLu5XceOYUa8IVAj53wQwmxEbzX8+VOR1KwIzpSVFVfUBVl6rq0szMzLEc+pT95s2DlNW28c3Li+0iEIecPSODSRNi+fZTpXT39TsdzqioKt/46y6S42P43AUznA7HAOkTYrlj5XReKK1lc4QsfehPZqoCfC+FzPc+Nm5UNXfykxf3sao4m1XFdiLLKdFRLq5YmMvBYx385s1DToczKn9+t4qNBxv52qVzrHYeQm45u4jM5Di+88zuiFimzp+EvgmYKSJTRSQWuBZYG9ywQsu315YAcPeV8xyOxMzKTmZVcTY/fXkflU3HnQ7HL83He/j+ut0sKUzjk9YmIqQkxkbzHx+ey/aKZh56+5DT4ZyyERO6qvYBdwLPAbuBR1W1RETuEZErAUTkDBGpBD4B/FJESoIZ9Fh6sbSW50tr+eLFM8lLS3A6HAN864piXCJ8+U/bw+Ko6ofPltHS2cv3PnoaLpd1VAw1Vy3K5bxZmfzoubKwn8boVzFYVdep6ixVna6q3/M+dpeqrvXe3qSq+ao6QVUnqWpEHMo2dfTwzSd3MSs7iVvPmep0OMYrf2Ii375yHhsPNfLL1/c7Hc4JbTjQwB83HuGWs4qYOznF6XDMEESE731kPm6Fu57chYbxerZ2dm8Yqsq//XkHx9q7ufeaRcRYe9yQ8rEleVx22mTufX4vu6panA5nSI0dPXxxzTaKJiXypVWznA7HnEBBeiJfXjWLF3fXsW5njdPhnDTLUsP4w4YjPF9ay7+tnsP8vFSnwzGDiAjf++h8MpLi+MKarSHXYtftVr762HYaO3q47/olJMXZdQuh7pazizgtL5VvPrmL6pbwLL1YQh9CWU0b3326lPNnZfIPZ1upJVSlJcby42sWcuhYB19asy2k6ukPvnmAl/fU8Y3L59oBQZiIjnLxk2sX0d3bz52PbA3L9UctoQ/S0tnL5x55l+T4GP77EwvtJFaIO3tGBt+4rJjnS2v5wfrdTocDwMaDjfzXs2VcOj+HG8+c4nQ4ZhSmZybxg48vYMvhJn64fo/T4YyafQ700dPn5o6HtnC4oYPf/8NyMpPjnA7J+OGWs4s41NDBr944SFHGBD613LkkWnq0lVt/t4nCSYn84OMLbOHwMHTFwlw2H2rkwTcPsrRoIqvnh896wXaE7qWqfO3PO3j7QAM//PgCVnjXtDShT0S46/JiVs7O5K4nS1jvUL+XQ8c6+PRvNpIUF81Dty63y/vD2L9fNpeF+al8+dHtbD3S5HQ4frOE7vU/L+zlL1ur+PKqWXxsSb7T4ZhRio5ycd/1S1hUkMbnHnmXRzdVjPyiAKpu6eSGX2+g3+3moVuX2TULYS4uOopffXqpp7PqbzexN0waeI37hK6q/PdzZfzfy+VcszSfz19ofTbClefIeBlnz8jgX/+8Y8wWMNhZ2cJH7n+L5uO9/PaWZczISh6TcU1wZaXE84dblxMb5eLGX2+gojH0r0we1wnd7Va+tbaE+14p57plBfznx6zmGe4SY6N58KalXHbaZL77zG6+/pcdQZ3SuG5nNZ/45d+Idrl47I4VLCxIC9pYZuwVTkrkoVuX09Xr5voH3+FAfbvTIZ3QuE3oXb39fOWx7fz+7cPcft40vv/R04iyGS0RIS46iv+7bjGfOX8aazZVcPlP3wz4xUcd3X189+lS/unhdymenMJfP3e2XQkaoWbnJPPQrcs43t3Px37+NzYeDN3OjOMyoR9u6ODjP/8bT2yt4quXzOLrl86xI/MIE+USvn7pXB6+dTkd3X189Gdv8f11u2lo7z7lfT9fUsOqe1/jwTcPct2yQh657UybERXhFuSn8cQ/nU36hFhueHADf90amg1nx920xWd3VfMvj+3A5RJ+fdNSLppr7XAj2VkzMnj2i+fxnWdK+dUbB3j4ncPcfHYRN60oIivF/wWae/rcPFdSw0NvH2bjoUZmZyfz+HWLWVqUHsToTSgpnJTIXz57Fp95aAtf+tM23iw/xl1XFJMSHzqzmcZNQq9q7uQ7T5XybEkNC/NTue/6JRSkJzodlhkDEyfEcu81i/inldP5yYv7uP+V/fzs1f0sLkhj9fwcTp+SzpRJiUyaEPveJzVVpaKxky1HGnn3cDPrd9VwrL2bgvQE7rq8mBtXTLH+PuNQWmIsD926nP97aR8/e7Wct/c38KOrF3DWjAynQwPGQULv6O7jt387xE9f3gfAv3xoNredO81WHRqHZmQlc9/1S/jnVe2s21HNsyU1fH/d368GTIqLJiE2is6efjp7+99rJTAhNoqzZmTwqeWFnDcz064eHudio1189UOzuWBOFl99bDvXP7iB1fNy+OqHZjMjK8nR2CI2oTd29PC7vx3id28fovl4L6vn5fDNK4ptfrBhemYSn79oJp+/aCaVTccpq2njcMNxjjQep7uvn4SYaBJiXeSmJbCkcCKzspPthLn5gNOnTOSZL5zDA68f4FevH+CF3bVcvSSffzx3KjOznZm6GlEJva/fzZvlx/jr1iqeLamhq9fNquJs7jh/OqdPmeh0eCYE5U9MJH+ild7MyUmMjeZLF8/ihjOncN/L5Tyy4Qh/2lzB8qnpfOrMKayam01CbNSYxRP2Cf1YezdvlR/j9b3HeG1vHcfae0hNiOHjS/K5+awix/5SGmPGj4ykOO6+ch6fv3AGj22p5OENh/nCH7cSF+3i3JmZXFKczYrpk8ifmBDUGXV+JXQRWQ38LxAFPKiqPxj0fBzwe+B0oAH4pKoeCmyoHvtq23htbz07KlvYWdXCwWMdAKQlxnDOjAyuWJjLytmZxEWP3V9FY4wBmJQUxx3nT+f2c6fxzoEGni+t5YXSWl7cXQtAdkocS6ekc8OZU4LSL2rEhC4iUcD9wCqgEtgkImtVtdRns1uBJlWdISLXAj8EPhnwaIHX9tbz3Wd2k5eWwPy8FK4+PZ9zZ2YwLzfV6pzGmJDgcglnzcjgrBkZfOuKYvbUtLH5UCObDjWx5XATH5qfE5Rx/TlCXwaUq+oBABFZA1wF+Cb0q4C7vbcfB+4TEdEgLM539en5fGRxHhlJdiGHMSb0iQhzJ6cwd3IKN64oAjxtR4LBn4SeB/i2rqsElg+3jar2iUgLMAk45ruRiNwO3O692y4iZScT9AlkDB7TfEDYv0efCv4QYf8ejYVP2fvkj2C8R8M2/B/Tk6Kq+gDwQLD2LyKbVXVpsPYfCew9Gpm9R/6x92lkY/0e+XN1TRVQ4HM/3/vYkNuISDSQiufkqDHGmDHiT0LfBMwUkakiEgtcC6wdtM1a4Cbv7auBl4NRPzfGGDO8EUsu3pr4ncBzeKYt/kZVS0TkHmCzqq4Ffg08JCLlQCOepO+EoJVzIoi9RyOz98g/9j6NbEzfI7EDaWOMiQzWocoYYyKEJXRjjIkQYZ/QRaRARF4RkVIRKRGRLzodUygSkXgR2Sgi273v07edjilUiUiUiGwVkaedjiUUicghEdkpIttEZLPT8YQqEUkTkcdFZI+I7BaRFcEeM+ybcwF9wFdU9V0RSQa2iMgLg1oTGOgGLlTVdhGJAd4UkfWq+o7TgYWgLwK7AVskdHgXqKpdVHRi/ws8q6pXe2cIBr2tZ9gfoatqtaq+673dhucXMc/ZqEKPegwsWR7j/bIz4oOISD5wGfCg07GY8CUiqcB5eGYAoqo9qtoc7HHDPqH7EpEiYDGwwdlIQpO3lLANqANeUFV7nz7oJ8C/Am6nAwlhCjwvIlu87TzMB00F6oH/5y3fPSgiE4I9aMQkdBFJAv4MfElVW52OJxSpar+qLsJzte8yEZnvdEyhREQuB+pUdYvTsYS4c1R1CXAp8DkROc/pgEJQNLAE+LmqLgY6gK8Fe9CISOjemvCfgYdV9S9OxxPqvB/9XgFWOx1LiDkbuFJEDgFrgAtF5A/OhhR6VLXK+28d8ASejqzm/SqBSp9PwY/jSfBBFfYJXTzLf/wa2K2q9zodT6gSkUwRSfPeTsDT337PiV81vqjq11U1X1WL8Fzt/LKq3uBwWCFFRCZ4Jx/gLSFcAuxyNqrQo6o1QIWIzPY+dBHvbzkeFJEwy+Vs4EZgp7c+DPDvqrrOwZhC0WTgd94FS1zAo6pq0/LMaGUDT3iXUYsGHlHVZ50NKWR9HnjYO8PlAHBLsAe0S/+NMSZChH3JxRhjjIcldGOMiRCW0I0xJkJYQjfGmAhhCd0YYyKEJXRjjIkQltDNuONt/5oxVq8b632a8csSuokIIhIJF8kZc0osoZuwICLfFJEyEXlTRP4oIl8VkVdF5CfeRRa+KCJXiMgGb3e7F0Uk2/vaSSLyvHdhjwcB8dnvDd6FP7aJyC+9V9L6E88HXicid4jIj3y2uVlE7juVcYwZDUvoJuSJyBnAx4GFeDr8LfV5OlZVl6rqj4E3gTO93e3W4GmDC/At4E1VnYenmVShd79zgU8CZ3u7UPYDn/IjnuFe92fgoz6bfhJYc7LjGDNa9jHVhIOzgSdVtQvoEpGnfJ77k8/tfOBPIjIZiAUOeh8/D/gYgKo+IyJN3scvAk4HNnl7kyTg6RU/kiFfp6r1InJARM4E9gFzgLeAz53kOMaMiiV0E+46fG7/FLhXVdeKyErg7hFeK8DvVPXroxzzRK9bA1yDp5PlE6qq3o6gJzOOMaNiJRcTDt4CrvAudJ0EXD7MdqlAlff2TT6Pvw5cDyAilwITvY+/BFwtIlne59JFZIof8ZzodU8AVwHX4UnupzKOMaNiCd2EPFXdBKwFdgDrgZ1AyxCb3g08JiJbAN8FjL8NnCciJXhKL0e8+y0FvoFnObUdwAt42gyPFM+wr1PVJjzr2k5R1Y2nMo4xo2Xtc01YEJEkVW0XkUQ8R9y3DywObozxsBq6CRcPiEgxEI+nHm3J3JhB7AjdmEFEZAMQN+jhG1V1pxPxGOMvS+jGGBMh7KSoMcZECEvoxhgTISyhG2NMhLCEbowxEeL/A686xT8xJq0aAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAEXCAYAAACjyo8UAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deZwcdZn48c8zPfd9ZjKZmWQm5L4hIcHlilwGXQkqKOAqKPvDXcVdf667urqLiLrqriv+XNldWXFFERHxihoJIEcAIWQSICdJJpNkjhxzTzL30c/vj6qBppmjJ9Mz1cfzfr36NdXf+lbVU101T1d/q+pboqoYY4yJXQleB2CMMWZqWaI3xpgYZ4neGGNinCV6Y4yJcZbojTEmxlmiN8aYGGeJPgKIyH+LyD+HaV6zRaRTRHzu+6dF5C/DMW93fn8QkZvDNb8JLPcrItIsIiene9nxSEQqRERFJNHrWCZDRG4Rkee8jsNrUb0Ro4GIHAWKgUFgCNgH/Ai4V1X9AKr6VxOY11+q6hOj1VHVWiBzclG/vrw7gXmq+hcB8786HPOeYByzgb8D5qhq43QvP1xERIH5qlrtdSwmvtgR/fR4t6pmAXOArwOfBe4L90Ki/ehrDLOBlulI8jH8GcY823ZjUFV7TeELOApcEVS2FvADy9z3PwS+4g4XAr8D2oFW4FmcL+Qfu9P0AJ3APwAVgAK3ArXA1oCyRHd+TwNfA14CTgO/AfLdceuB+pHiBTYA/cCAu7xXA+b3l+5wAvBPwDGgEeeXSo47bjiOm93YmoEvjPE55bjTN7nz+yd3/le46+x34/jhCNOuB+qBz7vLOQp8MGB8CvBNN45TwH8DaUHTfhY46X7OPndeh4EzwA6g3K2/CHjc3TYHgPcHLOeHwD3A793ptgHnuOO2up9Hl7seHxhnv9kIvOJus8PABrd8FrDJXX418H8CprkT+DnwgLv83cAC4B/d7VMHXBVQf6x9Y3j7JQZsn/uAE0AD8BX3c0p24/ykW88HPA/cMca6pbrbtNB9/wWcX7zZ7vsvA98ea79wx93iLutuoMWNqcD9fE676/Vl4Dm3vrh1G93xu3H/B2P95XkAsf5ihETvltcCf+0O/5A3Ev3XcBJRkvu6GJCR5hXwz/gjIANIG+Ef9Gn3H3OZW+cXwAPuuPWMkujd4TuH6waMf5o3Ev1HcZLNXJzmol8CPw6K7X/cuFYCfcDiUT6nH+Ekmix32oPAraPFGTTtejdRfAsnqV+Kk1AXuuPvdv/58935/xb4WtC033CnTQP+3k0CC93ksNJNIBk4yfIjOM2e5+J8sSwJ2I4tOF/kicBPgIcC4lScprDx9pm1QAdwJc6XXSmwyB23FfhPnGS5CicBXhawvXqBd7jL/xFwBCeRJgH/BzgStC1H2zeGt9/wfvQr4HtuvRk4SfRj7rhlQBuw2F3Wi4BvnHXcCrzPHX4M58vs6oBx7wlhv7jF3XafdNc3DXgIeNiNc5m7fsOJ/h04X9q57nZdDJR4nSOmJQ95HUCsvxg90b+Ie4TLmxP9Xe6O/ZaEEDyvgH/GuSOUBSb6rweMX4JzpO5j8on+j8DHA8YtxPkFkBgQR1nA+JeAG0ZYL58b05KAso8BT7vDb4kzaPr17j98RkDZw8A/u//QXbhH1u64t+EmPHfafiA1YPwBYOMIy/kA8GxQ2feALwZsx+8HjHsn8FrA+1AT/feAu0coL8c5z5MVUPY13F857vZ6PGDcu3F+Pfjc91luDLkh7Buv70c455j6cH8FuXVvBJ4KeP937ufWhnMeYrx1/DLwHXf+J4G/xWnWHD7aLwhhv7gFqA3ajwZwvxTdsn/hjUR/Gc4XxQW4vwri5WVt9N4pxfn5HezfcI6SHxORGhH5XAjzqpvA+GM4R3eFIUU5tlnu/ALnPZwYhgVeJdPNyCeKC92YgudVOoFY2lS1K2j6WUARkA7sEJF2EWkHHnXLhzWpam/A+3KcI8xgc4B1w/Nx5/VBYGZAnVDWdzyjLX8W0KqqZwLKgj+nUwHDPUCzqg4FvCcoplD2jTlu+YmA9f4ezpH9sPvdeptV9dBoKxbgGZwv2fNwfj09jvNL7AKgWlVbCG2/CIy/CGf/C14nAFT1SeC7OM1rjSJyr4hkhxBr1LNE7wEROR9nZ33LZV+qekZV/05V5wLXAJ8WkcuHR48yy9HKh5UHDM/GOeppxjnSTQ+Iy8ebE+B48z2O888dOO9B3pxsQtHsxhQ8r4YJzCNPRDKCpj/uzrsHWKqque4rR1UDk13wetYB54ywjDrgmYD55Kpqpqr+9QTiDMVoyz8O5ItIVkDZRD+nYKPtG8Hx9OG0qQ+vd7aqLg2o858455beISIXhbDcP+H8AnwPzme6z13+O3G+BCC0/SJw2zXh7H/B6/RGZdXvqOpqnF8vC3Ca6WKeJfppJCLZIvLnOO2ID6jq7hHq/LmIzBMRwWmnHcI5EQlOAp17Fov+CxFZIiLpOE1Dj7hHeQeBVBF5l4gk4ZzoSgmY7hRQISKj7Sc/Bf6viFSKSCbOz+SfqergRIJzY3kY+KqIZInIHODTOCcVJ+JLIpIsIhcDfw78XJ1LWP8HuFtEZgCISKmIvGOM+Xwf+LKIzBfHChEpwElkC0TkQyKS5L7OF5HFIcYX6va7D/iIiFwuIgluvItUtQ4nQX5NRFJFZAXOifiJfk6BRts3XqeqJ3Da0f/d3YcTROQcEbkUQEQ+BKzGaUr5G+B+d38Ylap247SXf4I3EvufgL8afj/R/cKt/0vgThFJF5ElOBcD4MZ5voisc/f1LpzzGf6R5hVrLNFPj9+KyBmcI6Mv4Jw0/MgodecDT+C0rb4A/KeqPuWO+xrwT+7P589MYPk/xmk/PonTBvo3AKraAXwcJ7E14Oz89QHT/dz92yIiO0eY7w/ceW/FOenXi3Ni7Gx80l1+Dc4vnQfd+YfqJE778HGck6B/paqvueM+i9Mc9qKInMb5fBeOMa9v4SSYx3CuzrgPp336DHAVcIO7nJO8cRI3FHfiJMF2EXn/aJVU9SWc/eNunC/7Z3jjqPZGnPbz4zgnSL+oY9xXEYIR940RfBjnCpt9OJ/zI0CJe4/Dt4EPq2qnqj4IVLmxj+cZnKaZlwLeZ+HsT8Mmul/cjtM0ddJdr/8NGJeN86XfhtOk04LTVBrzhq/mMCZqich6nF9IZV7HEk1E5Gmcz+37XsdippYd0RtjTIyzRG+MB0Tk826fRMGvP3gdW7i4/SKNtI6f9zq2eGNNN8YYE+PsiN4YY2JcxHUCVFhYqBUVFV6HYYwxUWXHjh3Nqlo00riIS/QVFRVUVVV5HYYxxkQVETk22jhrujHGmBhnid4YY2KcJXpjjIlxISV6EdkgIgdEpHqk3hRF5BIR2SkigyJyXdC42SLymIjsF5F9IlIRntCNMcaEYtxE7/ZoeA9wNU6Pbze6nQUFqsXp0OjBEWbxI+DfVHUxzgMVovaZn8YYE41CuepmLU7/0DUAIvIQzmPO9g1XUNWj7rg39QTnfiEkqurjbr3O8IRtjDEmVKE03ZTy5o786wn9gRALgHYR+aWIvCwi/+b+QngTEblNRKpEpKqpqSnEWRtjjAnFVJ+MTcR55ulngPNx+uK+JbiSqt6rqmtUdU1R0YjX+xtjjDlLoST6Bt78xJYyQn+iTT3wiqrWuA+j+DXOo8OMMcZMk1Da6LcD80WkEifB3wDcFOL8twO5IlKkqk04D+e1215jzIPbas962pvWzR6/kjFmUsY9onePxG8HtgD7gYdVda+I3CUi18Drj+iqB64Hvicie91ph3Cabf4oIrsBwXnCizHGmGkSUl83qroZ2BxUdkfA8HacJp2Rpn0cWDGJGI0xxkyC3RlrjDExzhK9McbEOEv0xhgT4yzRG2NMjLNEb4wxMc4SvTHGxDhL9MYYE+Ms0RtjTIyzRG+MMTHOEr0xxsQ4S/TGGBPjQurrxphhv331OLvq22nvHqCjZ4ArFhd7HZIxZhyW6E1IVJV/f+wg332qmtSkBPLSk0kQ4bF9p7h0QRFXLSlGRLwO0xgzAkv0Zlyqyld/v5/vP3eEG84v51/es5yEBGHIr9zxmz38ZFstp3sGeO95ZfgSLNkbE2ks0Ztxfem3+/jhn45yy59VcMefLyHBTea+BOEr1y7j1Ok+nth/CoDr15SPNStjjAcs0ZvXjfSkqAMnT3P/C8e48JwC5s/I5KHtdW+pc9miGQwM+XnmYBPrKvOZXZAxHeEaY0JkV92YUQ0O+fndrhMUZqbwjmUzx2yDX7+wiKyURDbvOYmqTmOUxpjxhJToRWSDiBwQkWoR+dwI4y8RkZ0iMigi140wPltE6kXku+EI2kyP5w+30NLVz7tXlJCYMPaukpLo44olxdS2drP3+OlpitAYE4pxE72I+IB7gKuBJcCNIrIkqFotcAvw4Ciz+TKw9ezDNNOto2eAp15rZHFJNvOLs0KaZvWcPIqzU3h070kG/f4pjtAYE6pQjujXAtWqWqOq/cBDwMbACqp6VFV3AW/57xaR1UAx8FgY4jXT5NE9J/Cr8q7lJSFPkyDC1ctKaO3q58Wa1imMzhgzEaEk+lIg8AxcvVs2LhFJAP4d+MzEQzNeOdHRw6v1HVw8v5D8jOQJTbugOIt5RZlsPdjEkN/a6o2JBFN9MvbjwGZVrR+rkojcJiJVIlLV1NQ0xSGZ8Txf3UyST7hoXtFZTf9n5xTQ2TfI/hPWVm9MJAgl0TcAgRdHl7lloXgbcLuIHAW+CXxYRL4eXElV71XVNaq6pqjo7JKLCY/TvQO8WtfB6jn5pCX7zmoe84uzyElLYvtRa74xJhKEch39dmC+iFTiJPgbgJtCmbmqfnB4WERuAdao6luu2jGR48XDLfhVufCcgrOehy9BWDMnjydfa6S1q3/CzT/GmPAa94heVQeB24EtwH7gYVXdKyJ3icg1ACJyvojUA9cD3xORvVMZtJka/YN+th1pZXFJNgWZKZOa15qKfACq7KjeGM+FdGesqm4GNgeV3REwvB2nSWesefwQ+OGEIzTTZmdtGz0DQ1w0r3DS88pJS2LhzCx2HGvj8sXF1geOMR6yO2MNAH6/8nx1M2V5acwpSA/LPNdW5HPGTsoa4zlL9AaA56qbaenq58JzCsPW3fCCmXZS1phIYIneAPBwVR1pST6WzsoO2zwTRDhvdh7VjZ2c6R0I23yNMRNjid7Q3t3PY/tOsao8l0RfeHeJFWU5KLC7oSOs8zXGhM4SvWHTq8fpH/Szek5e2OddnJ3KzOxUdtVbojfGK5boDQ9X1bGkJJtZuWlTMv+VZTnUtnbT1tU/JfM3xozNEn2c23f8NHsaTnP9mjGvjp2UFWW5AOyqb5+yZRhjRmeJPs79fEcdyb4Erl0VUj91ZyUvI5nyvDR2WTu9MZ6wRB/H+gf9/PrlBq5YMoO8Ke6mYGV5Lic6emk83TulyzHGvJUl+jj29IFG2roHuG711DXbDFtemoMAr9pJWWOmnSX6OPbbXSfIS0/i4vlT32NoVmoSc4sy2FXfbs+UNWaaWaKPUz39Q/xx/yk2LJtJUpivnR/NirJcWrr6OdFhzTfGTKeQOjUz0ePBbbUh1dvd0EF3/xDpyYkhTzNZS0uy+c0rDexu6JiySzmNMW9lR/Rxald9O5kpiVQWZkzbMtNTEjmnKJPdDR3WfGPMNLJEH4f6BoY4cPIMy0qzSQhTB2ahWl6aQ2tXPw3tPdO6XGPimSX6OLT/5GkG/cqK0txpX/bSWTn4RKzvG2OmkSX6OLSrvoPs1ERmh6nf+YlIS/Yxb4Y13xgznSzRx5me/iEOnepkeWnOtDfbDFtemkN79wD1bdZ8Y8x0CCnRi8gGETkgItUi8paHe4vIJSKyU0QGReS6gPJVIvKCiOwVkV0i8oFwBm8mbv/J0wypvt7/jBcWl2Rb840x02jcRC8iPuAe4GpgCXCjiCwJqlYL3AI8GFTeDXxYVZcCG4Bvi4h3Gcaw/8RpslMTKc3z7vLGtGQf84ud5hu/35pvjJlqoRzRrwWqVbVGVfuBh4CNgRVU9aiq7gL8QeUHVfWQO3wcaASm/jZMM6KBIT+HTnWyqGT6r7YJtqIsh46eAaqOtXkahzHxIJREXwrUBbyvd8smRETWAsnA4RHG3SYiVSJS1dTUNNFZmxAdbuqkf8jPkpLwPS7wbC0uySbJJ/z6lQavQzEm5k3LyVgRKQF+DHxEVf3B41X1XlVdo6priorsgH+q7Dt+mpTEBOZO401So0lJ9LG4JJvNu0/QP/iWXcIYE0ahJPoGoDzgfZlbFhIRyQZ+D3xBVV+cWHgmXPyqvHbyDAuKs8L+XNiztao8l/buAbYetF9xxkylUP7jtwPzRaRSRJKBG4BNoczcrf8r4Eeq+sjZh2kmq761m86+QRZHQLPNsPkzsshLT7LmG2Om2LiJXlUHgduBLcB+4GFV3Ssid4nINQAicr6I1APXA98Tkb3u5O8HLgFuEZFX3NeqKVkTM6Z9J86QILCwOMvrUF7nSxDetaKEJ/aforNv0OtwjIlZIfVeqaqbgc1BZXcEDG/HadIJnu4B4IFJxmjCYP+J08wtzCQt2ed1KG9y7apSHnixli17TvK+aXgAijHxKDIaa82UajrTR1NnH4tLIudofth5s/MozU3jN68e9zoUY2KWJfo4cODkaYCIap8flpAgbFw1i+cONdnzZI2ZIpbo48DBxk5mZKWQmz61DwA/W+89rwy/wi922klZY6aCPWEqxvUP+jnS3MXb5hZ4HcqIhp9uNSc/nfueqyE7NREJ8a7dm9bNnsrQjIkZdkQf4440dzLkVxZE0NU2I1lTkUdzZz/HWrq9DsWYmGOJPsYdPNVJkk+Y40Hf8xOxrDSH5MQE6/vGmClgiT7GHTx1hrmFmSRFyN2wo0lJ9LGiNIfdDe30Dgx5HY4xMSWy//vNpLR09tHS1c/84kyvQwnJmop8BoaU3fXWT70x4WSJPoYdauwEiPj2+WHleWnMyEqh6lir16EYE1Ms0cewQ6fOkJ+RTEFGZF5WGUxEWDMnj7q2Hk7aNfXGhI0l+hg16PdzuKmL+TMyQ75cMRKcOzuPxAThpSMtXodiTMywRB+jjrV00z/kj5pmm2EZKYksL83h5dp2+uykrDFhYYk+RlU3dpIgMLfI+4eMTNS6uQX0Dfp5pb7d61CMiQmW6GNUTVMn5XnppCRGVm+VoSjPS6MkJ5VtNa2o2sPDjZksS/QxqHdgiIb2HuYWRcdllcFEhAsqCzh5upfaVrtT1pjJskQfg462dOHX6Gy2GbayPJeUxAS2HbFLLY2ZLEv0MaimqYvEBGF2fmR3ezCW5MQEzpudx+6GDnv6lDGTFFKiF5ENInJARKpF5HMjjL9ERHaKyKCIXBc07mYROeS+bg5X4GZ0NU2dzM5Pj/huD8azrjKfIb+yw/q/MWZSxs0EIuID7gGuBpYAN4rIkqBqtcAtwINB0+YDXwTWAWuBL4pI3uTDNqPp7h/kREdvVDfbDJuRnUplYQYvHWnBbydljTlroRzyrQWqVbVGVfuBh4CNgRVU9aiq7gL8QdO+A3hcVVtVtQ14HNgQhrjNKI40d6HA3MLoPBEb7IK5BbR1D3Dw1BmvQzEmaoWS6EuBuoD39W5ZKEKaVkRuE5EqEalqamoKcdZmJDVNXST5hLL8NK9DCYslJdlkpSSyrcZOyhpztiKiEVdV71XVNaq6pqioyOtwotrhpk4qCjJITIiITTtpvgRhTUU+B0+dobWr3+twjIlKoWSDBqA84H2ZWxaKyUxrJqjpTB+NZ/qi9vr50aytzEcEXrJLLY05K6Ek+u3AfBGpFJFk4AZgU4jz3wJcJSJ57knYq9wyMwVerHE6AptbGP0nYgPlpCWxaGY2VcdaGRwKPg1kjBnPuIleVQeB23ES9H7gYVXdKyJ3icg1ACJyvojUA9cD3xORve60rcCXcb4stgN3uWVmCmw70kJKYgKzcmOjfT7Qurn5dPcPsff4aa9DMSbqJIZSSVU3A5uDyu4IGN6O0ywz0rQ/AH4wiRhNiLYfaWN2fjq+hOjpljhU5xRlkpeeRNWxVlaW53odjjFRJTbO2Bk6ugc4cOoMcwpiq9lmWIII583J43BTF212UtaYCbFEHyN21DotYhUF0dvtwXhWz85DgB21dqesMRNhiT5GbD/a5lw/nxe7iT43PZl5MzLZcazN7pQ1ZgIs0ceI7UdaWVaaQ3JibG/S1XPy6OgZ4LD74HNjzPhiOyvEid6BIXbVd3B+Rb7XoUy5JSXZpCX5qLKOzowJmSX6GLC7oYP+IX9cJPpEXwKrZuey78RpOylrTIgs0ceA7UedE7Gr58RHx6Br5uQx5Fc2vXrc61CMiQqW6GPA9iOtzJuRSX5GstehTIuSnDSKs1P4rSV6Y0JiiT7K+f1K1bE2zq+Ij6P5YSvLcqk61kZDe4/XoRgT8SzRR7mDjWc40zsYF+3zgZaX5gDYUb0xIbBEH+W2H3WuPom3RF+QmcLK8lxL9MaEwBJ9lKs62sqMrBTK8mKvI7PxXLNyFnuPn+Zwk11Tb8xYLNFHuZ21bayek4dI7HVkNp4/X1GCCGx6xY7qjRmLJfoo1niml7rWHs6bHV8nYocVZ6eyrjKf3756HLUuEYwZlSX6KLbzWDsA58XJ9fMjuWZlKTXNXdZPvTFjsEQfxV6ubSPZl8Cy0myvQ/HM1ctmkpgg/H73Ca9DMSZiWaKPYjtr21hamk1Kos/rUDyTl5HMurn5bNl70utQjIlYISV6EdkgIgdEpFpEPjfC+BQR+Zk7fpuIVLjlSSJyv4jsFpH9IvKP4Q0/fvUP+tlV3xG37fOB3rF0JjVNXVRbj5bGjGjcRC8iPuAe4GpgCXCjiCwJqnYr0Kaq84C7gW+45dcDKaq6HFgNfGz4S8BMzv4Tp+kb9FuiB65cUgxgR/XGjCKUI/q1QLWq1qhqP/AQsDGozkbgfnf4EeByca73UyBDRBKBNKAfsLNmYbDD7ab3vDn2/NSSnDRWluXw2L5TXodiTEQKJdGXAnUB7+vdshHrqOog0AEU4CT9LuAEUAt8U1VbgxcgIreJSJWIVDU1NU14JeLRzto2ZuWkUpITfzdKjeSqpTN5ta6dkx29XodiTMSZ6pOxa4EhYBZQCfydiMwNrqSq96rqGlVdU1RUNMUhxYaXa9s5N44vqwz2jqVO883j+6z5xphgoST6BqA84H2ZWzZiHbeZJgdoAW4CHlXVAVVtBJ4H1kw26Hh36nQvDe3xe6PUSM4pymRuYYY13xgzglAS/XZgvohUikgycAOwKajOJuBmd/g64El1blWsBS4DEJEM4ALgtXAEHs92DrfPz7b2+WEiwpVLi3nhcAsd3QNeh2NMRBk30btt7rcDW4D9wMOquldE7hKRa9xq9wEFIlINfBoYvgTzHiBTRPbifGH8r6ruCvdKxJsdx9pITkxg6awcr0OJKO9YOpNBv/LkATuqNyZQYiiVVHUzsDmo7I6A4V6cSymDp+scqdxMzst17SwvzSE50e53C7SqLJeirBSe2N/Ie84t8zocYyKGZYoo0z/oZ3dDhzXbjCAhQXj7wiK2HmhiYMjvdTjGRAxL9FFm/4nT9A/6OddOxI7oskUzONM3SJX7QBZjjCX6qPNyrZPAzrUj+hFdNL+IJJ/w1IFGr0MxJmJYoo8yO2vbmZltN0qNJjMlkbWV+Tz5miV6Y4aFdDLWTK8Ht9WOOu7ZQ03Myk0bs068u2xRMV/+3T7qWrspz0/3OhxjPGdH9FHkTO8Abd0DzLbkNabLFs0AsKN6Y1yW6KNIfVsPgCX6cVQWZlBZmGGJ3hiXJfooUtvaTYLArFxrnx/P2xfO4IWaFrr7B70OxRjPWaKPInWt3ZTkpJHks802nssWzaB/0M/z1S1eh2KM5yxjRAm/KvVtPXZyMURrK/PJSPZZ840xWKKPGqdO99I/5Gd2vjXbhCI5MYGL5hfyzIFGnP71jIlfluijRG1rNwDleXZEH6r1C2dwvKOXQ/YsWRPnLNFHibrWHjKSfeRnJHsdStRYv9B5iM1T1nxj4pwl+ihR69784zyK14SiJCeNRTOzePqAPZ7SxDdL9FGgu2+Q5s4+u37+LFy6sIiqY6109tllliZ+WaKPArVtTvv87AJL9BO1fsEMBoaU56ubvQ7FGM9Yoo8CtS3OjVJluZboJ2pNRR6ZKYk8bb1ZmjgWUqIXkQ0ickBEqkXkcyOMTxGRn7njt4lIRcC4FSLygojsFZHdIpIavvDjQ617o5Q9UWriknwJXDSvkKcPNNllliZujZs5RMSH8+zXq4ElwI0isiSo2q1Am6rOA+4GvuFOmwg8APyVqi4F1gP25OYJGPIrdW3WC+NkrF9YxImOXg6essssTXwK5RBxLVCtqjWq2g88BGwMqrMRuN8dfgS4XJzLQ64CdqnqqwCq2qKqQ+EJPT6cOt3LwJAyxxL9Wbt0+DJLa74xcSqURF8K1AW8r3fLRqyjqoNAB1AALABURLaIyE4R+YeRFiAit4lIlYhUNTXZpXCBjrXaidjJGr7M0q6nN/Fqqht9E4GLgA+6f98jIpcHV1LVe1V1jaquKSoqmuKQoktdazdZqYnkpiV5HUpUe/uiGew41sbpXms5NPEnlETfAJQHvC9zy0as47bL5wAtOEf/W1W1WVW7gc3AeZMNOp4ca+litt0oNWlvXziDQb/y/CG7zNLEn1AS/XZgvohUikgycAOwKajOJuBmd/g64El1LnHYAiwXkXT3C+BSYF94Qo999kSp8Dlvdi7ZqYnWm6WJS+M+M1ZVB0Xkdpyk7QN+oKp7ReQuoEpVNwH3AT8WkWqgFefLAFVtE5Fv4XxZKLBZVX8/ResSc4Y7MrMTsZOX6EvgkgVFPH2wCb9fSUiwX0gmfoT0cHBV3YzT7BJYdkfAcC9w/SjTPoBziaWZoNrWbnwJYk+UCpO3L5zB73adYN+J0ywrzfE6HGOmjd2BE8FqW7qZlZNKoj1RKiwuXViEiPVmaeKPZZAINTDkp769h4qCDK9DiRmFmSmsKMvlSbue3sSZkJpuzPSra+tmyK9UFFqiH82D22onPA0OISIAABlpSURBVE1hRjJPHmiktavf+vY3ccOO6CPU0eZuBOyIPswWzsxCFbYetBvzTPywRB+hjrZ0UZydSlqyz+tQYsqs3DQKM5PtMksTVyzRR6Ahv1Lb0k1FoV1WGW4JIly6YAbPHGxicMjvdTjGTAtL9BHoREcP/UN+a7aZIlcumUFHzwBVx9q8DsWYaWGJPgIdae4CsBOxU+Ti+UUk+xJ4Yt8pr0MxZlpYoo9AR1u6KchIJjvVOjKbChkpibztnAKe2H/KHkZi4oIl+gjj9ytHm7vsaH6KXbGkmKMt3Rxu6vI6FGOmnCX6CHOosZOegSFrn59ily+aAcAT+635xsQ+S/QR5qWjrQBU2hH9lJqVm8bSWdn80RK9iQOW6CPMS0dayU5NJC/d2uen2uWLi9lxrI3Wrn6vQzFmSlmijyCqyraaFioKM+xBI9PgysXF+NU6OTOxzxJ9BDnU2EnjmT7OKcr0OpS4sKw0m+LsFGunNzHPEn0Eec59zN28GZbop4OIcMXiYp452ETvwJDX4RgzZSzRR5DnqpupLMwgL916VZwuVy8robt/iGeskzMTw0JK9CKyQUQOiEi1iHxuhPEpIvIzd/w2EakIGj9bRDpF5DPhCTv29A/6ebGmhYvmFXodSlxZNzef3PQkHt1z0utQjJky4yZ6EfEB9wBXA0uAG0VkSVC1W4E2VZ0H3A18I2j8t4A/TD7c2PVybRvd/UNcNN8S/XRK8iVw5eJinth3ir5Ba74xsSmUI/q1QLWq1qhqP/AQsDGozkbgfnf4EeBycS8bEZFrgSPA3vCEHJueq24mQeCCuQVehxJ3rl4+kzN9g/ypusXrUIyZEqEk+lKgLuB9vVs2Yh1VHQQ6gAIRyQQ+C3xp8qHGtmcPNbOyPJecNLt+frpdOK+QrJRE/rDnhNehGDMlpvpk7J3A3araOVYlEblNRKpEpKqpKf5OinV0D7Crvp2LrX3eEymJPi5fPIPH9p1iwPqoNzEolETfAJQHvC9zy0asIyKJQA7QAqwD/lVEjgKfAj4vIrcHL0BV71XVNaq6pqioaMIrEe1eqGnGr3DR/Phb90ixYVkJ7d0DbKtp9ToUY8IulES/HZgvIpUikgzcAGwKqrMJuNkdvg54Uh0Xq2qFqlYA3wb+RVW/G6bYY8azh5rJSPZx7uxcr0OJW5cuKCItyWfNNyYmjZvo3Tb324EtwH7gYVXdKyJ3icg1brX7cNrkq4FPA2+5BNOM7rnqZi6YW0CSz25r8Epaso/LFs1gy96T9ohBE3MSQ6mkqpuBzUFldwQM9wLXjzOPO88ivphX3djJsZZubr2o0utQ4t67V5bw+90neP5wC5cusGY0EztCSvRm6jzuPs7uisXFHkcSPx7cVjti+cCQn9SkBO5+/CANbT0j1rlp3eypDM2YKWFtBR57bN9JlpfmMCs3zetQ4l6SL4HlpTnsO36a/kFrvjGxwxK9hxpP9/JKXTtXLbGj+UixqjyP/iE/+06c9joUY8LGEr2HntjfiCpcudQSfaSYU5BOTloSr9a1ex2KMWFjid5Dj+87yez8dBYWZ3kdinEliLCqPJdDjWfo7Bv0OhxjwsISvUc6+wZ5vrqFq5YU29OkIsyq8lz8Crvr7ajexAZL9B7ZerCJ/iE/V1r7fMQpzk6lJCeVl635xsQIS/QeeWzvSfIzklk9J8/rUMwIzi3Ppb6th1One70OxZhJs0Tvgf5BP0++1sjli2aQaHfDRqRzZ+fhSxBeOmJ935joZ1nGA08daOR07yDvXF7idShmFBkpiSwvzeHluja7pt5EPUv0HvjlznqKslK42J4mFdHOr8ind8DPLjspa6KcJfpp1tbVz5OvNXLtqlnWbBPhKgrSmZGVwktHrfnGRDfLNNNs06vHGRhS3ntemdehmHGICGsr86lv66GhfeS+b4yJBpbop9kvdtazpCSbxSXZXodiQnBueR5JPjspa6KbJfppdOjUGXbVd/C+1XY0Hy3Skn2sKM3l1bp2evqHvA7HmLNiiX4a/WJnA74E4ZqVs7wOxUzA284poH/Iz7YjLV6HYsxZsUQ/TYb8yq9fbmD9giKKslK8DsdMwKzcNBYUZ/J8dbMd1ZuoFFKiF5ENInJARKpF5C2PCRSRFBH5mTt+m4hUuOVXisgOEdnt/r0svOFHj8f3neLk6V6uX2PNNtHo0gUz6Oof4uGqOq9DMWbCxk30IuID7gGuBpYAN4rIkqBqtwJtqjoPuBv4hlveDLxbVZfjPDz8x+EKPNp8/9kayvLS7ElSUaqyMIM5+encu7WGAXumrIkyoRzRrwWqVbVGVfuBh4CNQXU2Ave7w48Al4uIqOrLqnrcLd8LpIlI3LVbvFzbRtWxNj56YaVdOx/FLl1YREN7D7955fj4lY2JIKFknVIg8PdqvVs2Yh1VHQQ6gIKgOu8Ddqpq39mFGr2+/9wRslITef/55V6HYiZhYXEWi2Zm8V9PV+P3q9fhGBOyaTm8FJGlOM05Hxtl/G0iUiUiVU1NTdMR0rSpa+3mD7tPcNO62WSm2LPYo5mIcPtl8zjc1MUvdtZ7HY4xIQsl0TcAgYeiZW7ZiHVEJBHIAVrc92XAr4APq+rhkRagqveq6hpVXVNUVDSxNYhw//v8URJEuOXPKrwOxYTBu5aXcO7sXP51ywF7ApWJGqEcYm4H5otIJU5CvwG4KajOJpyTrS8A1wFPqqqKSC7we+Bzqvp8+MKODvc9e4QHth1jWWkOT70WW79U4pWI8MV3L+Xae57nnqeq+eyGRV6HZMy4xj2id9vcbwe2APuBh1V1r4jcJSLXuNXuAwpEpBr4NDB8CebtwDzgDhF5xX3NCPtaRKhnDjYxMOi3XipjzKryXN57Xin3PXuE2pZur8MxZlwhNRqr6mZgc1DZHQHDvcD1I0z3FeArk4wxKtW3dfOnw82sKs+lJCfN63BMmH12wyIe3XOSr27ex/c+tMbrcIwZk13rN0X+/bGDAPZM2BhVnJ3KJ94+jy17T/HY3pNeh2PMmCzRT4E9DR386uUG/uycQnLTk70Ox0yRv7y4kmWl2fzDL3Zx3LoxNhHMEn2YqSr/snk/eelJrF8YW1cQmTdLSfTxHzeex8Cgn0899AqDdsesiVB2YXeYPbrnJH863MIX372ElESf1+GYKVZZmMFX37OcT/3sFb7zx0N8+qqFYV/Gg9tqz3ram9bNDmMkJlrZEX0YNZ3p4wu/3sPy0hz+4oI5Xodjpsm155Zy/eoy/uOpah7dY+31JvJYog8TVeXzv9pNZ98g33r/SpKsT5u48qWNS1lVnsvtD+60k7Mm4lg2CpNf7Gzg8X2n+PurFjK/OMvrcMw0S09O5P6PrmVZaQ6feHAnj+875XVIxrzOEn0Y1Ld186VNe1lbkc9HL6r0OhzjkezUJH5061qWzMrh4z/ZwY9fPIaqdX5mvGeJfpLO9A5w6w+rQOCb16/ElyBeh2Q8lJ2axI8+upYL5hbwz7/ew83/u52THb1hm79flc6+QRraezjc1EltazcnOnro6BmwLxUzKrvqZhIGhvx8/Cc7OdzUyQ8/spbZBeleh2SmWKhXwGxYOpO89GT+sOcE67/5FBfPL+Kr1y5jRnZqyMsaGPLz2okzvFjTQkN7Dw1tPTR19jE0ShfJGSmJlOWmUZ6fxrJZORNaloltlujPkqryxU17efZQM99433Iusv5sTAAR4YK5BcybkclvXz3O4/tO8dRrjVyxuJhLFhQxb0Ym82Zkkp7so2/QT9/AEA3tPRxq7OTQqTPsqu9gV30HPQPOM2rTk32U5aUxvziTnLQkctKSSE3yMTikDAz5OdM3SENbDw3t3Rzcf4Yn9jcyKyeVrr5B3r+mnJz0JI8/EeMlibSfe2vWrNGqqiqvwxiTqvKvWw7wX08f5uPrz+EfRunBcDLXP5vY0nymj87+QR7ZUU9rV/+YdZMTE1g8M4vz5uRx3uw8alu7yU1LQiS0ZsHTvQPsru/g1fp26tt6yEj2ccPa2Xz0okpKc63fpVglIjtUdcSOlyzRT9DgkJ8v/GoPP6uq48a15Xz12uUkjNIub4neBLpp3Wz8fuV4Rw/VjZ1UN3bSP+QnJdFHcmICxVkpLCjOojw//U3neiazH60qz+XerYf57a4TCPC+88q4/bJ5lOdbM2OsGSvRW9PNBPQODPHJn77M4/tO8cnL5vHpKxeEfJRlDEBCglCWl05ZXjrrF059j91LZmXz7RvO5e83LOJ/ttbw4Eu1/GJnPdevKecTbz+HsjxL+PHAEn2I9jR08OmHX+FQYyd3vnsJt1xol1Ga6FGam8ad1yzlry49h/98upqHXqrjkR11XLe6jI+vtyP8WGeJfhyDQ36+t7WGbz9xkLz0ZP73lvOn5UjMmHAYqdln0cxs/u+VC3jmYCMPV9Xzs+11rCzL5cJ5hcxy2/CjrY8c6w9obJboRzHkV36/+wTf+eMhqhs7edeKEr6ycRl5GdbtsIl+OWlJXLOylEsXzGDroSZ2HG3j5bp25hZlsK6ygL7BIeuUL4ZYog/S3t3PH/ac5L7njlDd2MmC4kz++y9Ws2HZTK9DMybsctKSePeKWVyxqJjtR1t5oaaFn75Uy+bdJ3j3yhLeubyENXPySU6Mvnsr/ap09AzQ2tVPe/cA7d39nOkbpLt/iJ7+QfoG/ajCT7Ydw5cgZKYkkpmSSE5aEiU5qczMSaMkN5U5+emU56dHdf9VISV6EdkA/D/AB3xfVb8eND4F+BGwGmgBPqCqR91x/wjcCgwBf6OqW8IWfRj4/crhpk6qjrXxxL5TbD3UxMCQsrA4i+/edC7vXFYy6lU1xsSKtGQflywo4qL5hVQ3dtLa1c/Pq+p54MVaMlMSuXBeAW+bW8CK8lyWlGSTmhQZR/uqSktXP8daumju7Ke5s+/1V0tnP4NBN5elJ/tIT04kPdlHWpKPBBFmZqcypEpX3yC1rd20dw/QeKaXwEl9CUJZXhqVhRnMLcxkblEGlYUZVBRmUJKdGvE5YtxELyI+4B7gSqAe2C4im1R1X0C1W4E2VZ0nIjcA3wA+ICJLgBuApcAs4AkRWaCqQ+FekZEMDPndb+8h2nv6ae3sp6Wrn+PtPRxt6eZYSxd7j5+mo2cAgJKcVD5yYSXXrJzF0lnZdkWNiTsJIiwozuKmdbPp7BvkT9XNPHWgiWcONLJlr9NRW2KCUFmYwZyCDCoK0pmVm0ZhVgqFGcnkZSSTmZJIRoqTTJN9CRNKggNDfvoG/fQODNHVN8iZXufV1t1Pa1c/LZ39nDrTy8mOXo6391DX2k1X/xvpJEEgPyOFwsxk5s/IojAzhfyMZPLSnZvMEkc4Kh+pjX5wyE9TZx8NbW/kiprmLmqaunixpoXegTceMpOcmEBpbhqluWnMyk2lKCuFwkznlZeeTHZaItmpSaSnOF8uqUm+af91EMoR/VqgWlVrAETkIWAjEJjoNwJ3usOPAN8VJ0tuBB5S1T7giIhUu/N7ITzhv6Gls493fHsr/YN+Bty7BYO/zQPlZyQzpyCdq5fNZPWcPFbPyaOyMMOSuzGuzJRErlo6k6uWzkRVOXm6l1frOthV3051YyfHWrp5rrrpTUlvJEk+IcmXQIIICeJcYqrqHI0rzvmwQb8yOORnjH/Z1+VnJFOSk0pZXhoXzC1gTkE6R5q7KMxIIS8jOSz9TSX6EijJSaMkJ401FflvGuf3KydO93KsuYsjLV0ca+l+vYuKpw800dLVP2o3FcMSBJJ8CST7EvD5BJ8IvgRheWkO991y/qTjf8v6hFCnFKgLeF8PrButjqoOikgHUOCWvxg0bWnwAkTkNuA2922niBwIKfqzUwg0HwNensKFeKQQaPY6iCkS9ev2wbFHR9T6jRPrRIV13cL9vzvJdQ3rum0HfvCRs5581KcdRcTJWFW9F7h3OpYlIlWj3T0W7Wzdolcsr5+tm/dCaShqAMoD3pe5ZSPWEZFEIAfnpGwo0xpjjJlCoST67cB8EakUkWSck6ubgupsAm52h68DnlSnE51NwA0ikiIilcB84KXwhG6MMSYU4zbduG3utwNbcC6v/IGq7hWRu4AqVd0E3Af82D3Z2orzZYBb72GcE7eDwCem64qbMUxLE5FHbN2iVyyvn62bxyKu90pjjDHhFb23ehljjAmJJXpjjIlxcZPoRWSDiBwQkWoR+ZzX8YSbiBwVkd0i8oqIRO6TW0IgIj8QkUYR2RNQli8ij4vIIfdvnpcxnq1R1u1OEWlwt90rIvJOL2M8WyJSLiJPicg+EdkrIn/rlsfKthtt/SJ++8VFG73bjcNBArpxAG4M6sYhqonIUWCNqkbMTTdnS0QuATqBH6nqMrfsX4FWVf26+0Wdp6qf9TLOszHKut0JdKrqN72MbbJEpAQoUdWdIpIF7ACuBW4hNrbdaOv3fiJ8+8XLEf3r3Tioaj8w3I2DiUCquhXn6q1AG4H73eH7cf7Bos4o6xYTVPWEqu50h88A+3HuhI+VbTfa+kW8eEn0I3XjEBUbaAIUeExEdrhdSsSaYlU94Q6fBIq9DGYK3C4iu9ymnahs2ggkIhXAucA2YnDbBa0fRPj2i5dEHw8uUtXzgKuBT7hNBDHJvRkvltoc/ws4B1gFnAD+3dtwJkdEMoFfAJ9S1dOB42Jh242wfhG//eIl0cd8Vwyq2uD+bQR+hdNcFUtOuW2kw22ljR7HEzaqekpVh1TVD/wPUbztRCQJJwn+RFV/6RbHzLYbaf2iYfvFS6IPpRuHqCUiGe7JIUQkA7gK2DP2VFEnsJuNm4HfeBhLWA0nQdd7iNJt53ZNfh+wX1W/FTAqJrbdaOsXDdsvLq66AXAvefo2b3Tj8FWPQwobEZmLcxQPTrcWD0bz+onIT4H1OF3AngK+CPwaeBiYjdNT7ftVNepOao6ybutxfvYrcBT4WECbdtQQkYuAZ4HdwHAn9Z/HaceOhW032vrdSIRvv7hJ9MYYE6/ipenGGGPiliV6Y4yJcZbojTEmxlmiN8aYGGeJ3hhjYpwlemOMiXGW6E3MEJFPiUi613EMc7uv/YzXcYxFRCoCu0w2sckSvfGUiIz73OIJ+BQQMYk+EoX58zZRwhK9mTT3qPA1EfmJiOwXkUdEJF1EVovIM26PmlsC+jt5WkS+7T4g5W9F5HwR+ZOIvCoiL4lIloj4ROTfRGS72yvgx9xp17vTPxKwTBGRvwFmAU+JyFNjxLpBRHa6y/qjW5YvIr92l/OiiKxwy+8UkftF5FkROSYi7xWRfxXnAS+Puv2eDD/0Zbj8JRGZN8Jyz3Gn2eHOb5GIJLrrt96t8zURGfGOZvcz+qU7vFFEekQkWURSRaTGLV/lxr9LRH413IviCJ/3anf9XwU+EbCMpW78r7jzmD+xPcFELFW1l70m9QIqcG7/vtB9/wPg74E/AUVu2Qdwup4AeBr4T3c4GagBznffZ+N043Ab8E9uWQpQBVTidBfQgdMxXQLwAk7PneDcfl44RpxFON1VV7rv892//wF80R2+DHjFHb4TeA5IAlYC3cDV7rhfAdcGLPcL7vCHgd8FTP8Zd/iPwHx3eB3wpDu8FKdf8yuAl4HkUWJPBGrc4W/i9N90IXAp8FO3fBdwqTt8F/Dt4M87oN4l7vC/AXsCPocPBmyXNK/3LXuF52U/40y41Knq8+7wAzh9gCwDHnf6gsKH04XrsJ+5fxcCJ1R1O4C63dqKyFXAChG5zq2XA8wH+oGXVLXerfcKzhfNcyHEeAGwVVWPuMsa7m/lIuB9btmTIlIgItnuuD+o6oCI7HbX4VG3fLe73GE/Dfh7d+BCxenW9s+An7ufBThfXqjqXhH5MfA74G3qPBjnLVR1UEQOi8hinN4RvwVc4sb0rIjkALmq+ow7yf3AzwNm8TM3lly33la3/Mc4XVuD86X5BREpA36pqodGisVEH0v0JlyCO006A+xV1beNUr9rnPkJ8ElV3fKmQqeZoy+gaIip3Y/7AFTVLyIDqjq8nv6g5eoow+D88mhX1VWjLGM50A7MGCeWrThJeQB4AvghTqL/+3Gmg/E/b1T1QRHZBrwL2CwiH1PVJ0OYt4lw1kZvwmW2iAwn9ZuAF4Gi4TIRSRKRpSNMdwAoEZHz3XpZ7gnDLcBfB7SDLxCnC+axnAGyxhj/InCJiFS688x3y58FPuiWrQeaNeiBGSH4QMDfFwJHuPM6IiLXu8sQEVnpDr8XyMc5Ov8P94h7NM/inHB+QVWbgAKcX0R7VLUDaBORi926HwKeCZ6BqrYD7eL0xAjueruxzMVpHvoOTlfCK0JdeRPZ7IjehMsBnCdb/QDYh9PeuwX4jtuskIjTTfTewIlUtV9EPoCT5NKAHpz26u/jNI3sFKe9o4nxnzV6L/CoiBxX1bcHj1TVJnEes/hLEUnAeQDGlTht6T8QkV047fA3B08bgjx3+j6cbmuDfRD4LxH5J5w2/4dEpAH4OnC5qtaJyHeB/zfG8rfhPIZvuNllFzAz4FfGzcB/i3OJaQ3wkVHm8xGc9VXgsYDy9wMfEpEBnEf+/ct4K22ig3VTbCZNnOdn/k5Vl3kciidE5CiwRlWbvY7FmJFY040xxsQ4O6I3Mck9qZgSVPwhVd3tRTwTISK/wrmUNNBng09MGxMqS/TGGBPjrOnGGGNinCV6Y4yJcZbojTEmxlmiN8aYGPf/AYTFWw/tfW0GAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visualize distribtutions for all potential features of API\n",
    "\n",
    "def feature_distplots(df):\n",
    "    \"\"\"Outputs distplots of all features in provided df.\"\"\"\n",
    "    for col, data in df.iteritems():\n",
    "        ax = sns.distplot(df[col])\n",
    "        plt.title(f\"Distribution of {col}\")\n",
    "        plt.show()\n",
    "\n",
    "feature_distplots(api_metrics.iloc[:,2:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create dictionary of hand rankings of 25 stories, provided by stakeholder\n",
    "rankings = {\"3130\": 1,  # key == Story ID, value == rank\n",
    "             \"3112\": 2,\n",
    "             \"5109\": 3,\n",
    "             \"3118\": 4,\n",
    "             \"3122\": 5,\n",
    "             \"3108\": 6,\n",
    "             \"5108\": 7,\n",
    "             \"3121\": 8,\n",
    "             \"3101\": 9,\n",
    "             \"3128\": 10,\n",
    "             \"3106\": 11,\n",
    "             \"3126\": 12,\n",
    "             \"3129\": 13,\n",
    "             \"3105\": 14,\n",
    "             \"5104\": 15,\n",
    "             \"5101\": 16,\n",
    "             \"5103\": 17,\n",
    "             \"3125\": 18,\n",
    "             \"3111\": 19,\n",
    "             \"3119\": 20,\n",
    "             \"3110\": 21,\n",
    "             \"3117\": 22,\n",
    "             \"3127\": 23,\n",
    "             \"3131\": 24,\n",
    "             \"3120\": 25} "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Subset dataframe for ranked human transcriptions\n",
    "ranked_stories_h = human_metrics[human_metrics[\"story_id\"].isin(rankings.keys())].reset_index(drop=True)\n",
    "# Add rankings\n",
    "ranked_stories_h[\"ranking\"] = ranked_stories_h[\"story_id\"].apply(lambda x: rankings[x])\n",
    "\n",
    "# Subset dataframe for ranked API transcriptions\n",
    "ranked_stories_api = api_metrics[api_metrics[\"story_id\"].isin(rankings.keys())].reset_index(drop=True)\n",
    "# Add rankings\n",
    "ranked_stories_api[\"ranking\"] = ranked_stories_api[\"story_id\"].apply(lambda x: rankings[x])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add Squad Score based on squad_score function\n",
    "ranked_stories_h[\"squad_score\"] = ranked_stories_h.iloc[:,:-1].apply(lambda x: squad_score(x, \"human\"), axis=1)\n",
    "# Add the rank based on the squad score\n",
    "ranked_stories_h[\"squad_score_rank\"] = ranked_stories_h[\"squad_score\"].rank(ascending=False)\n",
    "\n",
    "# Repeat for API transcriptions\n",
    "ranked_stories_api[\"squad_score\"] = ranked_stories_api.iloc[:,:-1].apply(lambda x: squad_score(x, \"api\"), axis=1)\n",
    "ranked_stories_api[\"squad_score_rank\"] = ranked_stories_api[\"squad_score\"].rank(ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "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>story_id</th>\n",
       "      <th>ranking</th>\n",
       "      <th>squad_score_rank</th>\n",
       "      <th>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3117</td>\n",
       "      <td>22</td>\n",
       "      <td>25.0</td>\n",
       "      <td>16.196703</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3105</td>\n",
       "      <td>14</td>\n",
       "      <td>12.0</td>\n",
       "      <td>37.525256</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3129</td>\n",
       "      <td>13</td>\n",
       "      <td>7.0</td>\n",
       "      <td>47.511298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3111</td>\n",
       "      <td>19</td>\n",
       "      <td>14.0</td>\n",
       "      <td>32.073730</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3118</td>\n",
       "      <td>4</td>\n",
       "      <td>13.0</td>\n",
       "      <td>35.631001</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  story_id  ranking  squad_score_rank  squad_score\n",
       "0     3117       22              25.0    16.196703\n",
       "1     3105       14              12.0    37.525256\n",
       "2     3129       13               7.0    47.511298\n",
       "3     3111       19              14.0    32.073730\n",
       "4     3118        4              13.0    35.631001"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Look at ranking outcome for human transcriptions\n",
    "ranked_stories_h[[\"story_id\", \"ranking\", \"squad_score_rank\", \"squad_score\"]].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "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>story_id</th>\n",
       "      <th>ranking</th>\n",
       "      <th>squad_score_rank</th>\n",
       "      <th>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3117</td>\n",
       "      <td>22</td>\n",
       "      <td>25.0</td>\n",
       "      <td>17.120997</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3105</td>\n",
       "      <td>14</td>\n",
       "      <td>11.0</td>\n",
       "      <td>34.042006</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3129</td>\n",
       "      <td>13</td>\n",
       "      <td>5.0</td>\n",
       "      <td>43.835181</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3111</td>\n",
       "      <td>19</td>\n",
       "      <td>14.0</td>\n",
       "      <td>31.448316</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3118</td>\n",
       "      <td>4</td>\n",
       "      <td>6.0</td>\n",
       "      <td>40.267507</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  story_id  ranking  squad_score_rank  squad_score\n",
       "0     3117       22              25.0    17.120997\n",
       "1     3105       14              11.0    34.042006\n",
       "2     3129       13               5.0    43.835181\n",
       "3     3111       19              14.0    31.448316\n",
       "4     3118        4               6.0    40.267507"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Look at ranking outcome for API transcriptions\n",
    "ranked_stories_api[[\"story_id\", \"ranking\", \"squad_score_rank\", \"squad_score\"]].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Human transcriptions:\n",
      "MAE: 4.8 rankings\n",
      "MSE: 38.0\n",
      "\n",
      "API transcriptions:\n",
      "MAE: 4.96 rankings\n",
      "MSE: 39.68\n"
     ]
    }
   ],
   "source": [
    "# Calculate MAE & MSE for difference between \n",
    "# stakeholder ranking and ranking generated by Squad Score\n",
    "y_true_h = ranked_stories_h[\"ranking\"]\n",
    "y_pred_h = ranked_stories_h[\"squad_score_rank\"]\n",
    "mae_h = mae(y_true_h, y_pred_h)\n",
    "mse_h = mse(y_true_h, y_pred_h)\n",
    "print(\"Human transcriptions:\")\n",
    "print(f\"MAE: {mae_h} rankings\")\n",
    "print(f\"MSE: {mse_h}\")\n",
    "\n",
    "y_true_api = ranked_stories_api[\"ranking\"]\n",
    "y_pred_api = ranked_stories_api[\"squad_score_rank\"]\n",
    "mae_api = mae(y_true_api, y_pred_api)\n",
    "mse_api = mse(y_true_api, y_pred_api)\n",
    "print(\"\\nAPI transcriptions:\")\n",
    "print(f\"MAE: {mae_api} rankings\")\n",
    "print(f\"MSE: {mse_api}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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>grade_level</th>\n",
       "      <th>story_length</th>\n",
       "      <th>avg_word_len</th>\n",
       "      <th>quotes_num</th>\n",
       "      <th>unique_words_num</th>\n",
       "      <th>adj_num</th>\n",
       "      <th>percent_complex_words</th>\n",
       "      <th>ranking</th>\n",
       "      <th>squad_score</th>\n",
       "      <th>squad_score_rank</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>grade_level</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.133356</td>\n",
       "      <td>0.106430</td>\n",
       "      <td>0.099504</td>\n",
       "      <td>0.121204</td>\n",
       "      <td>-0.019996</td>\n",
       "      <td>0.102992</td>\n",
       "      <td>-0.097073</td>\n",
       "      <td>0.126602</td>\n",
       "      <td>-0.249615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>story_length</th>\n",
       "      <td>0.133356</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.170846</td>\n",
       "      <td>0.288018</td>\n",
       "      <td>0.985773</td>\n",
       "      <td>0.699843</td>\n",
       "      <td>0.161942</td>\n",
       "      <td>-0.547640</td>\n",
       "      <td>0.877277</td>\n",
       "      <td>-0.837752</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>avg_word_len</th>\n",
       "      <td>0.106430</td>\n",
       "      <td>0.170846</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.280723</td>\n",
       "      <td>0.179889</td>\n",
       "      <td>0.147636</td>\n",
       "      <td>0.483906</td>\n",
       "      <td>-0.318545</td>\n",
       "      <td>0.496456</td>\n",
       "      <td>-0.469200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>quotes_num</th>\n",
       "      <td>0.099504</td>\n",
       "      <td>0.288018</td>\n",
       "      <td>0.280723</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.301507</td>\n",
       "      <td>0.324317</td>\n",
       "      <td>0.199650</td>\n",
       "      <td>-0.339448</td>\n",
       "      <td>0.579847</td>\n",
       "      <td>-0.506412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>unique_words_num</th>\n",
       "      <td>0.121204</td>\n",
       "      <td>0.985773</td>\n",
       "      <td>0.179889</td>\n",
       "      <td>0.301507</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.749681</td>\n",
       "      <td>0.222655</td>\n",
       "      <td>-0.536605</td>\n",
       "      <td>0.894578</td>\n",
       "      <td>-0.852126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>adj_num</th>\n",
       "      <td>-0.019996</td>\n",
       "      <td>0.699843</td>\n",
       "      <td>0.147636</td>\n",
       "      <td>0.324317</td>\n",
       "      <td>0.749681</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.161838</td>\n",
       "      <td>-0.578159</td>\n",
       "      <td>0.789394</td>\n",
       "      <td>-0.657188</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>percent_complex_words</th>\n",
       "      <td>0.102992</td>\n",
       "      <td>0.161942</td>\n",
       "      <td>0.483906</td>\n",
       "      <td>0.199650</td>\n",
       "      <td>0.222655</td>\n",
       "      <td>0.161838</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.185505</td>\n",
       "      <td>0.340646</td>\n",
       "      <td>-0.328446</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ranking</th>\n",
       "      <td>-0.097073</td>\n",
       "      <td>-0.547640</td>\n",
       "      <td>-0.318545</td>\n",
       "      <td>-0.339448</td>\n",
       "      <td>-0.536605</td>\n",
       "      <td>-0.578159</td>\n",
       "      <td>-0.185505</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.634993</td>\n",
       "      <td>0.618462</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>squad_score</th>\n",
       "      <td>0.126602</td>\n",
       "      <td>0.877277</td>\n",
       "      <td>0.496456</td>\n",
       "      <td>0.579847</td>\n",
       "      <td>0.894578</td>\n",
       "      <td>0.789394</td>\n",
       "      <td>0.340646</td>\n",
       "      <td>-0.634993</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.918517</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>squad_score_rank</th>\n",
       "      <td>-0.249615</td>\n",
       "      <td>-0.837752</td>\n",
       "      <td>-0.469200</td>\n",
       "      <td>-0.506412</td>\n",
       "      <td>-0.852126</td>\n",
       "      <td>-0.657188</td>\n",
       "      <td>-0.328446</td>\n",
       "      <td>0.618462</td>\n",
       "      <td>-0.918517</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       grade_level  story_length  avg_word_len  quotes_num  \\\n",
       "grade_level               1.000000      0.133356      0.106430    0.099504   \n",
       "story_length              0.133356      1.000000      0.170846    0.288018   \n",
       "avg_word_len              0.106430      0.170846      1.000000    0.280723   \n",
       "quotes_num                0.099504      0.288018      0.280723    1.000000   \n",
       "unique_words_num          0.121204      0.985773      0.179889    0.301507   \n",
       "adj_num                  -0.019996      0.699843      0.147636    0.324317   \n",
       "percent_complex_words     0.102992      0.161942      0.483906    0.199650   \n",
       "ranking                  -0.097073     -0.547640     -0.318545   -0.339448   \n",
       "squad_score               0.126602      0.877277      0.496456    0.579847   \n",
       "squad_score_rank         -0.249615     -0.837752     -0.469200   -0.506412   \n",
       "\n",
       "                       unique_words_num   adj_num  percent_complex_words  \\\n",
       "grade_level                    0.121204 -0.019996               0.102992   \n",
       "story_length                   0.985773  0.699843               0.161942   \n",
       "avg_word_len                   0.179889  0.147636               0.483906   \n",
       "quotes_num                     0.301507  0.324317               0.199650   \n",
       "unique_words_num               1.000000  0.749681               0.222655   \n",
       "adj_num                        0.749681  1.000000               0.161838   \n",
       "percent_complex_words          0.222655  0.161838               1.000000   \n",
       "ranking                       -0.536605 -0.578159              -0.185505   \n",
       "squad_score                    0.894578  0.789394               0.340646   \n",
       "squad_score_rank              -0.852126 -0.657188              -0.328446   \n",
       "\n",
       "                        ranking  squad_score  squad_score_rank  \n",
       "grade_level           -0.097073     0.126602         -0.249615  \n",
       "story_length          -0.547640     0.877277         -0.837752  \n",
       "avg_word_len          -0.318545     0.496456         -0.469200  \n",
       "quotes_num            -0.339448     0.579847         -0.506412  \n",
       "unique_words_num      -0.536605     0.894578         -0.852126  \n",
       "adj_num               -0.578159     0.789394         -0.657188  \n",
       "percent_complex_words -0.185505     0.340646         -0.328446  \n",
       "ranking                1.000000    -0.634993          0.618462  \n",
       "squad_score           -0.634993     1.000000         -0.918517  \n",
       "squad_score_rank       0.618462    -0.918517          1.000000  "
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Look at correlations\n",
    "# Note that the correlation between Squad Score and stakeholder ranking is -.60\n",
    "ranked_stories_api.iloc[:,2:].corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate complexity ratings from all textstat complexity metrics \n",
    "# For use in comparing to Squad Score correlations\n",
    "\n",
    "def get_textstat_metrics(transcription):\n",
    "    \"\"\"\n",
    "    Return 8 complexity metrics from texstat package for a transcription.\n",
    "    \n",
    "    Metrics:\n",
    "    Flesch-Kincaid Grade Level, Fog Scale, SMOG Index, \n",
    "    Automated Readability Index, Coleman-Liau Index, Linsear Write Formula,\n",
    "    Dale-Chall Readability Score, and a compiled \"Readabilty Consensus\"\n",
    "    \"\"\"\n",
    "    \n",
    "    row = []\n",
    "    \n",
    "    metric_functions = [\"smog_index\", \"flesch_kincaid_grade\", \"coleman_liau_index\", \n",
    "                   \"automated_readability_index\", \"dale_chall_readability_score\",\n",
    "                   \"linsear_write_formula\", \"gunning_fog\", \"text_standard\"]\n",
    "    \n",
    "    # Add all textstat functions to row \n",
    "    for metric in metric_functions:\n",
    "        row.append(getattr(textstat, metric)(transcription))\n",
    "    \n",
    "    return row"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generate new dataframe\n",
    "metric_functions = [\"smog_index\", \"flesch_kincaid_grade\", \"coleman_liau_index\", \n",
    "                    \"automated_readability_index\", \"dale_chall_readability_score\",\n",
    "                    \"linsear_write_formula\", \"gunning_fog\", \"text_standard\"]\n",
    "\n",
    "textstat_output = ranked_stories_api[\"transcription\"].apply(get_textstat_metrics)\n",
    "\n",
    "textstat_df = pd.DataFrame.from_records(textstat_output, \n",
    "                          index=ranked_stories_api[\"story_id\"],\n",
    "                          columns=metric_functions)\n",
    "\n",
    "# Add rankings to df\n",
    "textstat_df[\"ranking\"] = ranked_stories_api[\"ranking\"].tolist()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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>smog_index</th>\n",
       "      <th>flesch_kincaid_grade</th>\n",
       "      <th>coleman_liau_index</th>\n",
       "      <th>automated_readability_index</th>\n",
       "      <th>dale_chall_readability_score</th>\n",
       "      <th>linsear_write_formula</th>\n",
       "      <th>gunning_fog</th>\n",
       "      <th>ranking</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>smog_index</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.611097</td>\n",
       "      <td>0.076767</td>\n",
       "      <td>-0.615090</td>\n",
       "      <td>-0.420572</td>\n",
       "      <td>-0.565149</td>\n",
       "      <td>-0.605051</td>\n",
       "      <td>-0.323771</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>flesch_kincaid_grade</th>\n",
       "      <td>-0.611097</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.213118</td>\n",
       "      <td>0.999333</td>\n",
       "      <td>0.659239</td>\n",
       "      <td>0.913174</td>\n",
       "      <td>0.998816</td>\n",
       "      <td>0.063661</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>coleman_liau_index</th>\n",
       "      <td>0.076767</td>\n",
       "      <td>0.213118</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.231851</td>\n",
       "      <td>0.343967</td>\n",
       "      <td>0.235261</td>\n",
       "      <td>0.183021</td>\n",
       "      <td>-0.195179</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>automated_readability_index</th>\n",
       "      <td>-0.615090</td>\n",
       "      <td>0.999333</td>\n",
       "      <td>0.231851</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.660066</td>\n",
       "      <td>0.910650</td>\n",
       "      <td>0.997681</td>\n",
       "      <td>0.062233</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>dale_chall_readability_score</th>\n",
       "      <td>-0.420572</td>\n",
       "      <td>0.659239</td>\n",
       "      <td>0.343967</td>\n",
       "      <td>0.660066</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.420005</td>\n",
       "      <td>0.645076</td>\n",
       "      <td>-0.007742</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>linsear_write_formula</th>\n",
       "      <td>-0.565149</td>\n",
       "      <td>0.913174</td>\n",
       "      <td>0.235261</td>\n",
       "      <td>0.910650</td>\n",
       "      <td>0.420005</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.913920</td>\n",
       "      <td>0.106353</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>gunning_fog</th>\n",
       "      <td>-0.605051</td>\n",
       "      <td>0.998816</td>\n",
       "      <td>0.183021</td>\n",
       "      <td>0.997681</td>\n",
       "      <td>0.645076</td>\n",
       "      <td>0.913920</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.066445</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>ranking</th>\n",
       "      <td>-0.323771</td>\n",
       "      <td>0.063661</td>\n",
       "      <td>-0.195179</td>\n",
       "      <td>0.062233</td>\n",
       "      <td>-0.007742</td>\n",
       "      <td>0.106353</td>\n",
       "      <td>0.066445</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                              smog_index  flesch_kincaid_grade  \\\n",
       "smog_index                      1.000000             -0.611097   \n",
       "flesch_kincaid_grade           -0.611097              1.000000   \n",
       "coleman_liau_index              0.076767              0.213118   \n",
       "automated_readability_index    -0.615090              0.999333   \n",
       "dale_chall_readability_score   -0.420572              0.659239   \n",
       "linsear_write_formula          -0.565149              0.913174   \n",
       "gunning_fog                    -0.605051              0.998816   \n",
       "ranking                        -0.323771              0.063661   \n",
       "\n",
       "                              coleman_liau_index  automated_readability_index  \\\n",
       "smog_index                              0.076767                    -0.615090   \n",
       "flesch_kincaid_grade                    0.213118                     0.999333   \n",
       "coleman_liau_index                      1.000000                     0.231851   \n",
       "automated_readability_index             0.231851                     1.000000   \n",
       "dale_chall_readability_score            0.343967                     0.660066   \n",
       "linsear_write_formula                   0.235261                     0.910650   \n",
       "gunning_fog                             0.183021                     0.997681   \n",
       "ranking                                -0.195179                     0.062233   \n",
       "\n",
       "                              dale_chall_readability_score  \\\n",
       "smog_index                                       -0.420572   \n",
       "flesch_kincaid_grade                              0.659239   \n",
       "coleman_liau_index                                0.343967   \n",
       "automated_readability_index                       0.660066   \n",
       "dale_chall_readability_score                      1.000000   \n",
       "linsear_write_formula                             0.420005   \n",
       "gunning_fog                                       0.645076   \n",
       "ranking                                          -0.007742   \n",
       "\n",
       "                              linsear_write_formula  gunning_fog   ranking  \n",
       "smog_index                                -0.565149    -0.605051 -0.323771  \n",
       "flesch_kincaid_grade                       0.913174     0.998816  0.063661  \n",
       "coleman_liau_index                         0.235261     0.183021 -0.195179  \n",
       "automated_readability_index                0.910650     0.997681  0.062233  \n",
       "dale_chall_readability_score               0.420005     0.645076 -0.007742  \n",
       "linsear_write_formula                      1.000000     0.913920  0.106353  \n",
       "gunning_fog                                0.913920     1.000000  0.066445  \n",
       "ranking                                    0.106353     0.066445  1.000000  "
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Look at correlations\n",
    "# Note that best correlation to ranking is smog_index, with only -.32\n",
    "# Because of this, decision was made that Squad Score is a better fit for our use case\n",
    "# than any textstat package\n",
    "textstat_df.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "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>story_id</th>\n",
       "      <th>story_length</th>\n",
       "      <th>avg_word_len</th>\n",
       "      <th>quotes_num</th>\n",
       "      <th>unique_words_num</th>\n",
       "      <th>adj_num</th>\n",
       "      <th>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3132</td>\n",
       "      <td>1375</td>\n",
       "      <td>5.092593</td>\n",
       "      <td>6</td>\n",
       "      <td>138</td>\n",
       "      <td>15</td>\n",
       "      <td>47.843668</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  story_id  story_length  avg_word_len  quotes_num  unique_words_num  adj_num  \\\n",
       "0     3132          1375      5.092593           6               138       15   \n",
       "\n",
       "   squad_score  \n",
       "0    47.843668  "
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Generate CSV with Squad Score and metrics used for Squad Score\n",
    "csv_features = scaled_features.copy()\n",
    "csv_features.insert(0, \"story_id\")\n",
    "squad_score_metrics = api_metrics[csv_features].merge(api_squad_score)\n",
    "\n",
    "squad_score_metrics.head(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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>story_id</th>\n",
       "      <th>story_length</th>\n",
       "      <th>avg_word_len</th>\n",
       "      <th>quotes_num</th>\n",
       "      <th>unique_words_num</th>\n",
       "      <th>adj_num</th>\n",
       "      <th>squad_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>3229</td>\n",
       "      <td>296</td>\n",
       "      <td>4.169014</td>\n",
       "      <td>0</td>\n",
       "      <td>32</td>\n",
       "      <td>2</td>\n",
       "      <td>1.668999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>66</th>\n",
       "      <td>3240</td>\n",
       "      <td>173</td>\n",
       "      <td>5.088235</td>\n",
       "      <td>0</td>\n",
       "      <td>26</td>\n",
       "      <td>3</td>\n",
       "      <td>14.401436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3117</td>\n",
       "      <td>439</td>\n",
       "      <td>4.877778</td>\n",
       "      <td>1</td>\n",
       "      <td>56</td>\n",
       "      <td>3</td>\n",
       "      <td>17.120997</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>3202</td>\n",
       "      <td>535</td>\n",
       "      <td>4.652174</td>\n",
       "      <td>3</td>\n",
       "      <td>64</td>\n",
       "      <td>7</td>\n",
       "      <td>19.418121</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>3127</td>\n",
       "      <td>548</td>\n",
       "      <td>4.849558</td>\n",
       "      <td>2</td>\n",
       "      <td>56</td>\n",
       "      <td>6</td>\n",
       "      <td>20.389726</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>132</th>\n",
       "      <td>5219</td>\n",
       "      <td>2148</td>\n",
       "      <td>5.290640</td>\n",
       "      <td>42</td>\n",
       "      <td>222</td>\n",
       "      <td>17</td>\n",
       "      <td>91.975460</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>78</th>\n",
       "      <td>5254</td>\n",
       "      <td>3467</td>\n",
       "      <td>5.317485</td>\n",
       "      <td>4</td>\n",
       "      <td>354</td>\n",
       "      <td>26</td>\n",
       "      <td>95.950853</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>102</th>\n",
       "      <td>5213</td>\n",
       "      <td>2928</td>\n",
       "      <td>5.030928</td>\n",
       "      <td>34</td>\n",
       "      <td>299</td>\n",
       "      <td>29</td>\n",
       "      <td>104.660013</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>108</th>\n",
       "      <td>5235</td>\n",
       "      <td>3207</td>\n",
       "      <td>4.995327</td>\n",
       "      <td>32</td>\n",
       "      <td>302</td>\n",
       "      <td>32</td>\n",
       "      <td>107.548101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>5234</td>\n",
       "      <td>2989</td>\n",
       "      <td>5.555762</td>\n",
       "      <td>26</td>\n",
       "      <td>295</td>\n",
       "      <td>47</td>\n",
       "      <td>119.110260</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>167 rows × 7 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    story_id  story_length  avg_word_len  quotes_num  unique_words_num  \\\n",
       "34      3229           296      4.169014           0                32   \n",
       "66      3240           173      5.088235           0                26   \n",
       "3       3117           439      4.877778           1                56   \n",
       "51      3202           535      4.652174           3                64   \n",
       "10      3127           548      4.849558           2                56   \n",
       "..       ...           ...           ...         ...               ...   \n",
       "132     5219          2148      5.290640          42               222   \n",
       "78      5254          3467      5.317485           4               354   \n",
       "102     5213          2928      5.030928          34               299   \n",
       "108     5235          3207      4.995327          32               302   \n",
       "117     5234          2989      5.555762          26               295   \n",
       "\n",
       "     adj_num  squad_score  \n",
       "34         2     1.668999  \n",
       "66         3    14.401436  \n",
       "3          3    17.120997  \n",
       "51         7    19.418121  \n",
       "10         6    20.389726  \n",
       "..       ...          ...  \n",
       "132       17    91.975460  \n",
       "78        26    95.950853  \n",
       "102       29   104.660013  \n",
       "108       32   107.548101  \n",
       "117       47   119.110260  \n",
       "\n",
       "[167 rows x 7 columns]"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "squad_score_metrics.sort_values(\"squad_score\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [],
   "source": [
    "squad_score_metrics.to_csv(\"squad_score_metrics.csv\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "StorySquad (Python 3.8)",
   "language": "python",
   "name": "storysquad"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}