NHS_ARD_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "c8284331",
   "metadata": {},
   "source": [
    "# Detecting Acute Respiratory Distress in NHS 111 Calls\n",
    "\n",
    "## Table of Contents\n",
    "<ul>\n",
    "<li><a href=\"#intro\">Introduction</a></li>\n",
    "<li><a href=\"#wrangling\">Data Wrangling</a></li>\n",
    "<li><a href=\"#eda\">Exploratory Data Analysis</a></li>\n",
    "<li><a href=\"#modelling\">Predictive Modelling</a></li>\n",
    "<li><a href=\"#conclusions\">Conclusions & Next Steps</a></li>\n",
    "</ul>"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "103ab786",
   "metadata": {},
   "source": [
    "<a id='intro'></a>\n",
    "## Introduction"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "15196990",
   "metadata": {},
   "source": [
    "This notebook performs data analysis and modelling using a (private) dataset containing information on 74 calls made to NHS 111, with callers labelled as being in acute respiratory distress (ARD, 40) or not (non-ARD, 34).\n",
    "\n",
    "For privacy reasons, the content and metadata of the conversations were completely removed:\n",
    "- No audio\n",
    "- No transcription\n",
    "- No time of day or location\n",
    "- No caller identity data\n",
    "\n",
    "The dataset consists of 74 CSV files, each containing a subset of the IBM Watson STT JSON response. Each row in the CSV corresponds to one word in the call, and there are five features per word:\n",
    "\n",
    "1. **TimeFrom:** Beginning time for the word\n",
    "2. **TimeTo:** End time for the word\n",
    "3. **Speaker:** Who is saying the word (0: operator, 1: caller)\n",
    "4. **Gap between:** A derived feature: the pause between the previous speaker finishing and the current speaker starting. \n",
    "5. **Confidence:** IBM's confidence that the transcription it gave to the word is correct (a probability)\n",
    "\n",
    "The question, therefore, is as follows: *To what extent, with this limited dataset, is it possible to classify callers as being in ARD or not?*\n",
    "\n",
    "The motivation for asking this question is to enable NHS 111 to automatically prioritise individuals in ARD so that their waiting times are minimised."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "id": "4d0a05b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Basic libraries\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "from scipy.stats import kurtosis\n",
    "\n",
    "# Model selection & pipelining \n",
    "from sklearn.model_selection import cross_val_score, cross_val_predict, StratifiedKFold, GridSearchCV\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import make_pipeline\n",
    "\n",
    "# Models\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from xgboost import XGBClassifier\n",
    "\n",
    "# Metrics\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix, classification_report, roc_curve\n",
    "\n",
    "sns.set_theme()\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a17aa9d5",
   "metadata": {},
   "source": [
    "### Helper functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "063f2a0b",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculates number of turns in the conversation for both speakers\n",
    "\n",
    "def calc_num_turns(arr):\n",
    "    num_operator = 0\n",
    "    num_caller = 0\n",
    "    curr = -1\n",
    "    \n",
    "    for i in arr:\n",
    "        if i != curr:\n",
    "            curr = i\n",
    "            if i == 0:\n",
    "                num_operator += 1\n",
    "            else: \n",
    "                num_caller += 1\n",
    "                \n",
    "    return [num_operator, num_caller]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "id": "a811f7d1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculates the kurtosis of turn length for both speakers\n",
    "\n",
    "def calc_turn_kurtosis(df):\n",
    "    operator_turn_lens = []\n",
    "    caller_turn_lens = []\n",
    "    curr = 0\n",
    "    time_from = 0\n",
    "    \n",
    "    for i in range(df.shape[0]):\n",
    "        if i == (df.shape[0] - 1):\n",
    "            if curr == 0:\n",
    "                operator_turn_lens.append(df['TimeTo'][i] - time_from)\n",
    "            else: \n",
    "                caller_turn_lens.append(df['TimeTo'][i] - time_from)\n",
    "        elif df['Speaker'][i + 1] != curr:\n",
    "            if curr == 0:\n",
    "                operator_turn_lens.append(df['TimeTo'][i] - time_from)\n",
    "            else: \n",
    "                caller_turn_lens.append(df['TimeTo'][i] - time_from)\n",
    "            curr = df['Speaker'][i + 1]\n",
    "            time_from = df['TimeFrom'][i + 1]\n",
    "            \n",
    "    return [kurtosis(np.array(operator_turn_lens)), kurtosis(np.array(caller_turn_lens))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "id": "bcc89026",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generic bar chart plotting\n",
    "\n",
    "def plot_bar_chart(axis, xlabels, metrics, ylabel):\n",
    "    accuracy = pd.DataFrame(data=metrics, columns=[ylabel], index=xlabels)\n",
    "    ax = sns.barplot(ax=axis, x=xlabels, y=ylabel, data=accuracy)\n",
    "    for p in ax.patches:\n",
    "        ax.annotate(format(p.get_height(), '.2f'), \n",
    "                       (p.get_x() + p.get_width() / 2., p.get_height()), \n",
    "                       ha = 'center', va = 'center', \n",
    "                       xytext = (0, 9), \n",
    "                       textcoords = 'offset points');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "id": "12d49937",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Feature analysis plotting\n",
    "\n",
    "def plot_feature_analysis(title, colour, non_ARD_avg, ARD_avg):\n",
    "    fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(14,5))\n",
    "    fig.suptitle(title, fontweight='bold')\n",
    "    \n",
    "    # Scatter plot\n",
    "    df.plot.scatter(ax=ax1, x=title, y='in_ARD', c=colour)\n",
    "    \n",
    "    # Bar chart \n",
    "    plot_bar_chart(ax2, xlabels=['Not in ARD', 'In ARD'], metrics=np.array([non_ARD_avg, ARD_avg]), ylabel='Averages');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "id": "6b70d730",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Confusion matrix plotting \n",
    "\n",
    "def plot_conf_matrix(axis, conf_matrix):\n",
    "    sns.heatmap(conf_matrix, ax=axis, annot=True, cmap=\"YlGnBu\", fmt='g')    \n",
    "    axis.set_ylabel('Actual label')\n",
    "    axis.set_xlabel('Predicted label')\n",
    "    axis.set_title('Confusion Matrix');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "9f3b8b48",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Specificity vs. sensitivity plotting \n",
    "\n",
    "def plot_spec_sens(y_probs, spec, sens, title):\n",
    "    fpr, tpr, _ = roc_curve(y, y_probs)\n",
    "    fig, axis = plt.subplots(figsize=(7,7))\n",
    "    axis.plot(1 - fpr, tpr)\n",
    "    plt.scatter(spec, sens, color='red', alpha=1, zorder=3)\n",
    "    axis.set_xlabel('Specificity (TNR)')\n",
    "    axis.set_ylabel('Sensitivity (TPR)')\n",
    "    plt.title(title, fontweight='bold');"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "49f881e6",
   "metadata": {},
   "source": [
    "<a id='wrangling'></a>\n",
    "## Data Wrangling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d434755",
   "metadata": {},
   "source": [
    "### Loading files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "6febcd62",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set relative path to non_ARD and ARD dir\n",
    "\n",
    "non_ARD_dir = 'dummy_data/non_ARD'\n",
    "ARD_dir = 'dummy_data/ARD'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "31abf84f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from os import listdir\n",
    "\n",
    "non_ARD_files = [f'dummy_data/non_ARD/{f}' for f in listdir(non_ARD_dir)]\n",
    "ARD_files = [f'dummy_data/ARD/{f}' for f in listdir(ARD_dir)]\n",
    "\n",
    "files = {\n",
    "    0: non_ARD_files,\n",
    "    1: ARD_files\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "id": "4e363463",
   "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>TimeFrom</th>\n",
       "      <th>TimeTo</th>\n",
       "      <th>Speaker</th>\n",
       "      <th>Gap between speakers</th>\n",
       "      <th>Confidence</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.00</td>\n",
       "      <td>0.62</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.49</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.65</td>\n",
       "      <td>0.82</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.82</td>\n",
       "      <td>0.91</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.91</td>\n",
       "      <td>1.11</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1.11</td>\n",
       "      <td>1.22</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.67</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   TimeFrom  TimeTo  Speaker  Gap between speakers  Confidence\n",
       "0      0.00    0.62        1                   0.0        0.49\n",
       "1      0.65    0.82        1                   0.0        0.67\n",
       "2      0.82    0.91        1                   0.0        0.67\n",
       "3      0.91    1.11        1                   0.0        0.67\n",
       "4      1.11    1.22        1                   0.0        0.67"
      ]
     },
     "execution_count": 131,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Preview one file\n",
    "\n",
    "df_preview = pd.read_csv(ARD_files[0], delimiter=',')\n",
    "df_preview.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3c2da0c",
   "metadata": {},
   "source": [
    "### Creating features from raw data\n",
    "\n",
    "Four features were identified as being of interest for the caller:\n",
    "\n",
    "- **Words per minute:** Individuals in ARD might speak more slowly (although they could also be panicking and speaking more quickly).\n",
    "- **Words per turn:** Individuals in ARD might be able to say fewer words per turn.\n",
    "- **% of conversation speaking:** Individuals in ARD might say less overall in the conversation.\n",
    "- **Kurtosis of turn length:** Kurtosis is a measure of a distribution's 'tailedness', or the propensity for the distribution to produce outliers. Individuals in ARD, given the limitations this places on their speaking, might have a lower kurtosis of turn length due to the physical limitations they are experiencing.\n",
    "\n",
    "These were also calculated for the operator, since it might be that we need to account for these when looking at caller features. For example, individuals have a tendency to mirror others, so if the operator naturally speaks more quickly then we might observe callers (on average) speaking more quickly with that operator too.\n",
    "\n",
    "Such 'operator' features were not analysed here, however, both for brevity and to avoid the *curse of dimensionality* (we should avoid having too many features with only 74 data points). This is left as an avenue for future work. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "id": "c9264ae9",
   "metadata": {},
   "outputs": [],
   "source": [
    "dataset = []\n",
    "\n",
    "# For each output class (0 = non-ARD, 1 = ARD)\n",
    "for output_class in [0, 1]:\n",
    "    for path in files[output_class]:\n",
    "        \n",
    "        df = pd.read_csv(path, delimiter=',')\n",
    "        \n",
    "        row = []\n",
    "                \n",
    "        ## RAW DF WRANGLING\n",
    "        # Fix speaker mislabelling (e.g. 1 and 2 instead of 0 and 1)\n",
    "        if df['Speaker'].isin([1, 2]).all():\n",
    "            df['Speaker'] = df['Speaker'] - 1\n",
    "        elif df['Speaker'].isin([0, 2]).all():\n",
    "            df.loc[df['Speaker'] == 2, 'Speaker'] = 1\n",
    "            \n",
    "        # Add derived column\n",
    "        df['duration'] = df['TimeTo'] - df['TimeFrom']\n",
    "        \n",
    "        \n",
    "        # Helper masks\n",
    "        operator_mask = df['Speaker'] == 0\n",
    "        caller_mask = df['Speaker'] == 1\n",
    "        \n",
    "        \n",
    "        ## 1) WORDS PER MINUTE\n",
    "        operator_words = df[operator_mask].shape[0]\n",
    "        operator_duration = df[operator_mask]['duration'].sum()\n",
    "        operator_wp_min = operator_words / operator_duration * 60\n",
    "        row.append(operator_wp_min)\n",
    "        \n",
    "        caller_words = df[caller_mask].shape[0]\n",
    "        caller_duration = df[caller_mask]['duration'].sum()\n",
    "        caller_wp_min = caller_words / caller_duration * 60\n",
    "        row.append(caller_wp_min)\n",
    "        \n",
    "        \n",
    "        ## 2) WORDS PER TURN\n",
    "        operator_turns, caller_turns = calc_num_turns(df['Speaker'])\n",
    "        operator_wp_turn = operator_words / operator_turns\n",
    "        row.append(operator_wp_turn)\n",
    "        caller_wp_turn = caller_words / caller_turns\n",
    "        row.append(caller_wp_turn)\n",
    "        \n",
    "        \n",
    "        ## 3) % CONVERSATION DURATION\n",
    "        operator_percent = operator_duration / (caller_duration + operator_duration)\n",
    "        row.append(operator_percent)\n",
    "        caller_percent = 1 - operator_percent\n",
    "        row.append(caller_percent)\n",
    "        \n",
    "        \n",
    "        ## 4) KURTOSIS TURN LENGTH\n",
    "        operator_len_kurt, caller_len_kurt = calc_turn_kurtosis(df)\n",
    "        row.append(operator_len_kurt)\n",
    "        row.append(caller_len_kurt)\n",
    "        \n",
    "        # Output label\n",
    "        row.append(output_class)\n",
    "        \n",
    "        dataset.append(row)\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "id": "fdb57258",
   "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>operator_wp_min</th>\n",
       "      <th>caller_wp_min</th>\n",
       "      <th>operator_wp_turn</th>\n",
       "      <th>caller_wp_turn</th>\n",
       "      <th>operator_percent</th>\n",
       "      <th>caller_percent</th>\n",
       "      <th>operator_turn_kurt</th>\n",
       "      <th>caller_turn_kurt</th>\n",
       "      <th>in_ARD</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>228.029242</td>\n",
       "      <td>192.445820</td>\n",
       "      <td>17.878049</td>\n",
       "      <td>6.317073</td>\n",
       "      <td>0.704883</td>\n",
       "      <td>0.295117</td>\n",
       "      <td>4.073135</td>\n",
       "      <td>12.685567</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>232.144276</td>\n",
       "      <td>182.647601</td>\n",
       "      <td>12.815789</td>\n",
       "      <td>9.131579</td>\n",
       "      <td>0.524764</td>\n",
       "      <td>0.475236</td>\n",
       "      <td>3.454322</td>\n",
       "      <td>8.030907</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>217.118595</td>\n",
       "      <td>232.330827</td>\n",
       "      <td>41.666667</td>\n",
       "      <td>12.875000</td>\n",
       "      <td>0.795746</td>\n",
       "      <td>0.204254</td>\n",
       "      <td>0.942607</td>\n",
       "      <td>-1.064707</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>233.180794</td>\n",
       "      <td>195.127557</td>\n",
       "      <td>13.547619</td>\n",
       "      <td>6.738095</td>\n",
       "      <td>0.627212</td>\n",
       "      <td>0.372788</td>\n",
       "      <td>4.345672</td>\n",
       "      <td>7.257747</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>218.520756</td>\n",
       "      <td>179.834134</td>\n",
       "      <td>17.583333</td>\n",
       "      <td>8.956522</td>\n",
       "      <td>0.627681</td>\n",
       "      <td>0.372319</td>\n",
       "      <td>1.798440</td>\n",
       "      <td>6.767570</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   operator_wp_min  caller_wp_min  operator_wp_turn  caller_wp_turn  \\\n",
       "0       228.029242     192.445820         17.878049        6.317073   \n",
       "1       232.144276     182.647601         12.815789        9.131579   \n",
       "2       217.118595     232.330827         41.666667       12.875000   \n",
       "3       233.180794     195.127557         13.547619        6.738095   \n",
       "4       218.520756     179.834134         17.583333        8.956522   \n",
       "\n",
       "   operator_percent  caller_percent  operator_turn_kurt  caller_turn_kurt  \\\n",
       "0          0.704883        0.295117            4.073135         12.685567   \n",
       "1          0.524764        0.475236            3.454322          8.030907   \n",
       "2          0.795746        0.204254            0.942607         -1.064707   \n",
       "3          0.627212        0.372788            4.345672          7.257747   \n",
       "4          0.627681        0.372319            1.798440          6.767570   \n",
       "\n",
       "   in_ARD  \n",
       "0       0  \n",
       "1       0  \n",
       "2       0  \n",
       "3       0  \n",
       "4       0  "
      ]
     },
     "execution_count": 133,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Convert dataset to DataFrame & preview\n",
    "\n",
    "columns = [\n",
    "    'operator_wp_min', \n",
    "    'caller_wp_min',\n",
    "    'operator_wp_turn', \n",
    "    'caller_wp_turn',\n",
    "    'operator_percent',\n",
    "    'caller_percent',\n",
    "    'operator_turn_kurt',\n",
    "    'caller_turn_kurt',\n",
    "    'in_ARD'\n",
    "]\n",
    "\n",
    "df = pd.DataFrame(dataset, columns=columns)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd5b8e35",
   "metadata": {},
   "source": [
    "<a id='eda'></a>\n",
    "## Exploratory Data Analysis"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "734cb66d",
   "metadata": {},
   "source": [
    "### Summary statistics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "77d0cb3e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Num. nulls:\t 0\n",
      "Num. duplicates: 0\n"
     ]
    }
   ],
   "source": [
    "# Count nulls and duplicates\n",
    "\n",
    "num_nulls = len(df.index) - len(df.dropna().index)\n",
    "num_duplicates = sum(df.duplicated())\n",
    "\n",
    "print('Num. nulls:\\t', num_nulls)\n",
    "print('Num. duplicates:', num_duplicates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "id": "90be3fac",
   "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>caller_wp_min</th>\n",
       "      <th>caller_wp_turn</th>\n",
       "      <th>caller_percent</th>\n",
       "      <th>caller_turn_kurt</th>\n",
       "      <th>in_ARD</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>74.000000</td>\n",
       "      <td>74.000000</td>\n",
       "      <td>74.000000</td>\n",
       "      <td>74.000000</td>\n",
       "      <td>74.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>201.213950</td>\n",
       "      <td>8.205312</td>\n",
       "      <td>0.394206</td>\n",
       "      <td>8.004148</td>\n",
       "      <td>0.540541</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>21.172603</td>\n",
       "      <td>3.853361</td>\n",
       "      <td>0.113929</td>\n",
       "      <td>7.207496</td>\n",
       "      <td>0.501756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>154.838710</td>\n",
       "      <td>2.844444</td>\n",
       "      <td>0.204254</td>\n",
       "      <td>-1.500000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>187.096680</td>\n",
       "      <td>6.049291</td>\n",
       "      <td>0.318129</td>\n",
       "      <td>2.531612</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>200.514116</td>\n",
       "      <td>7.287356</td>\n",
       "      <td>0.362504</td>\n",
       "      <td>6.224098</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>213.339148</td>\n",
       "      <td>9.212243</td>\n",
       "      <td>0.444907</td>\n",
       "      <td>12.154385</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>277.842324</td>\n",
       "      <td>22.250000</td>\n",
       "      <td>0.788993</td>\n",
       "      <td>27.385155</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       caller_wp_min  caller_wp_turn  caller_percent  caller_turn_kurt  \\\n",
       "count      74.000000       74.000000       74.000000         74.000000   \n",
       "mean      201.213950        8.205312        0.394206          8.004148   \n",
       "std        21.172603        3.853361        0.113929          7.207496   \n",
       "min       154.838710        2.844444        0.204254         -1.500000   \n",
       "25%       187.096680        6.049291        0.318129          2.531612   \n",
       "50%       200.514116        7.287356        0.362504          6.224098   \n",
       "75%       213.339148        9.212243        0.444907         12.154385   \n",
       "max       277.842324       22.250000        0.788993         27.385155   \n",
       "\n",
       "          in_ARD  \n",
       "count  74.000000  \n",
       "mean    0.540541  \n",
       "std     0.501756  \n",
       "min     0.000000  \n",
       "25%     0.000000  \n",
       "50%     1.000000  \n",
       "75%     1.000000  \n",
       "max     1.000000  "
      ]
     },
     "execution_count": 135,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Drop operator features (not used in this analysis, as discussed previously)\n",
    "\n",
    "df.drop(['operator_wp_min', 'operator_wp_turn', 'operator_percent', 'operator_turn_kurt'], axis=1, inplace=True)\n",
    "df.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "id": "fb4de08c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "caller_wp_min       74\n",
       "caller_wp_turn      74\n",
       "caller_percent      74\n",
       "caller_turn_kurt    74\n",
       "in_ARD               2\n",
       "dtype: int64"
      ]
     },
     "execution_count": 136,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Number of unique values\n",
    "\n",
    "df.nunique()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2f3fd79f",
   "metadata": {},
   "source": [
    "### Relationships between features\n",
    "\n",
    "The correlations between many of the features below suggest that *Naive Bayes* should not be used, since its conditional independence assumption would be violated."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "id": "7e8f5b17",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 25 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Overall summary of relationships between variables\n",
    "\n",
    "pd.plotting.scatter_matrix(df, figsize=(10, 10));"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "96d4d782",
   "metadata": {},
   "source": [
    "### Investigating feature relevance\n",
    "\n",
    "- **caller_wp_min**: Appears to be irrelevant. Callers' baseline (when not in ARD) speaking rates would potentially be needed here, since different people naturally have different speaking rates.  \n",
    "- **caller_wp_turn**: Appears to be highly relevant\n",
    "- **caller_percent**: Appears to be relevant, but in a non-linear fashion: non-ARD callers are less likely to have extreme values\n",
    "- **caller_turn_kurt**: Appears to be highly relevant"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "id": "bddacfe1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Helper mask\n",
    "ARD_mask = df['in_ARD'] == 1\n",
    "\n",
    "# caller_wp_min\n",
    "non_ARD_wp_min = df[~ARD_mask]['caller_wp_min'].mean()\n",
    "ARD_wp_min = df[ARD_mask]['caller_wp_min'].mean()\n",
    "\n",
    "# caller_wp_turn\n",
    "non_ARD_wp_turn = df[~ARD_mask]['caller_wp_turn'].mean()\n",
    "ARD_wp_turn = df[ARD_mask]['caller_wp_turn'].mean()\n",
    "\n",
    "# caller_percent\n",
    "non_ARD_percent = df[~ARD_mask]['caller_percent'].mean()\n",
    "ARD_percent = df[ARD_mask]['caller_percent'].mean()\n",
    "\n",
    "# caller_turn_kurt\n",
    "non_ARD_turn_kurt = df[~ARD_mask]['caller_turn_kurt'].mean()\n",
    "ARD_turn_kurt = df[ARD_mask]['caller_turn_kurt'].mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "id": "11a28d19",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# caller_wp_min analysis\n",
    "\n",
    "plot_feature_analysis('caller_wp_min', 'red', non_ARD_wp_min, ARD_wp_min)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "id": "95cc2fef",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# caller_wp_turn analysis\n",
    "\n",
    "plot_feature_analysis('caller_wp_turn', 'blue', non_ARD_wp_turn, ARD_wp_turn)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "id": "f97bfd84",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# caller_percent analysis\n",
    "\n",
    "plot_feature_analysis('caller_percent', 'green', non_ARD_percent, ARD_percent)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "0b4a5462",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# caller_turn_kurt analysis\n",
    "\n",
    "plot_feature_analysis('caller_turn_kurt', 'purple', non_ARD_turn_kurt, ARD_turn_kurt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "id": "821d223e",
   "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>caller_wp_turn</th>\n",
       "      <th>caller_percent</th>\n",
       "      <th>caller_turn_kurt</th>\n",
       "      <th>in_ARD</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6.317073</td>\n",
       "      <td>0.295117</td>\n",
       "      <td>12.685567</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9.131579</td>\n",
       "      <td>0.475236</td>\n",
       "      <td>8.030907</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>12.875000</td>\n",
       "      <td>0.204254</td>\n",
       "      <td>-1.064707</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6.738095</td>\n",
       "      <td>0.372788</td>\n",
       "      <td>7.257747</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8.956522</td>\n",
       "      <td>0.372319</td>\n",
       "      <td>6.767570</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   caller_wp_turn  caller_percent  caller_turn_kurt  in_ARD\n",
       "0        6.317073        0.295117         12.685567       0\n",
       "1        9.131579        0.475236          8.030907       0\n",
       "2       12.875000        0.204254         -1.064707       0\n",
       "3        6.738095        0.372788          7.257747       0\n",
       "4        8.956522        0.372319          6.767570       0"
      ]
     },
     "execution_count": 143,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Drop caller_wp_min since it appears to be irrelevant\n",
    "\n",
    "df.drop(['caller_wp_min'], axis=1, inplace=True)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9348ad69",
   "metadata": {},
   "source": [
    "<a id='modelling'></a>\n",
    "## Predictive Modelling"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b9c39073",
   "metadata": {},
   "source": [
    "### Model scoring\n",
    "\n",
    "The following algorithms were fit to the data:\n",
    "\n",
    "- Logistic Regression\n",
    "- K-Nearest Neighbours (KNN)\n",
    "- Support Vector Machine\n",
    "- Decision Tree\n",
    "- XGBoost\n",
    "\n",
    "Nested cross-validation was employed to both enable hyperparameter tuning and to increase confidence in estimates of test set performance (due to the limited data available). \n",
    "\n",
    "Models were scored according to their accuracy (% of predictions correct), since this metric is applicable to all of the algorithms (other metrics, such as AUC, are not). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "id": "37669a4e",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create design and output matrices\n",
    "X = df.iloc[:,:-1]\n",
    "y = df.iloc[:,-1:]\n",
    "\n",
    "# Cross-validation techniques\n",
    "inner_cv = StratifiedKFold(n_splits=4, shuffle=True, random_state=44)\n",
    "outer_cv = StratifiedKFold(n_splits=4, shuffle=True, random_state=44)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "id": "67df26c1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Logistic Regression\n",
    "\n",
    "lr = LogisticRegression()\n",
    "lr_pipe = make_pipeline(StandardScaler(), lr)\n",
    "\n",
    "param_grid = {\n",
    "    'logisticregression__C': np.logspace(-4, 4, 8)\n",
    "}\n",
    "\n",
    "clf = GridSearchCV(lr_pipe, param_grid, cv=inner_cv)\n",
    "accuracies = cross_val_score(clf, X, y.values.ravel(), cv=outer_cv)\n",
    "\n",
    "lr_score = accuracies.mean()\n",
    "\n",
    "# Additional statistics\n",
    "lr_pred = cross_val_predict(clf, X, y.values.ravel(), cv=outer_cv)\n",
    "lr_pred_prob = cross_val_predict(clf, X, y.values.ravel(), cv=outer_cv, method='predict_proba')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "id": "bb19ddf6",
   "metadata": {},
   "outputs": [],
   "source": [
    "# KNN\n",
    "\n",
    "neigh = KNeighborsClassifier()\n",
    "neigh_pipe = make_pipeline(StandardScaler(), neigh)\n",
    "\n",
    "param_grid = {\n",
    "    'kneighborsclassifier__n_neighbors': range(1, 10, 1)\n",
    "}\n",
    "\n",
    "clf = GridSearchCV(neigh_pipe, param_grid, cv=inner_cv)\n",
    "accuracies = cross_val_score(clf, X, y.values.ravel(), cv=outer_cv)\n",
    "\n",
    "neigh_score = accuracies.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "id": "7e031acc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Support Vector Machine\n",
    "\n",
    "svc = SVC()\n",
    "svc_pipe = make_pipeline(StandardScaler(), svc)\n",
    "\n",
    "param_grid = [\n",
    "  {'svc__C': [0.01, 0.1, 1, 10, 100], 'svc__kernel': ['linear']},\n",
    "  {'svc__C': [0.01, 0.1, 1, 10, 100], 'svc__gamma': [0.001, 0.01, 0.1, 1], 'svc__kernel': ['rbf']}\n",
    "]\n",
    "\n",
    "clf = GridSearchCV(svc_pipe, param_grid, cv=inner_cv)\n",
    "accuracies = cross_val_score(clf, X, y.values.ravel(), cv=outer_cv)\n",
    "\n",
    "svc_score = accuracies.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "id": "6956effc",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Decision Tree\n",
    "\n",
    "dt = DecisionTreeClassifier()\n",
    "dt_pipe = make_pipeline(StandardScaler(), dt)\n",
    "\n",
    "param_grid = {\n",
    "    'decisiontreeclassifier__max_depth': range(3, 16, 2),\n",
    "    'decisiontreeclassifier__criterion': ['gini', 'entropy']\n",
    "}\n",
    "\n",
    "clf = GridSearchCV(dt_pipe, param_grid, cv=inner_cv)\n",
    "accuracies = cross_val_score(clf, X, y.values.ravel(), cv=outer_cv)\n",
    "\n",
    "dt_score = accuracies.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "id": "8ea65d38",
   "metadata": {},
   "outputs": [],
   "source": [
    "# XGBoost\n",
    "\n",
    "xgb = XGBClassifier(objective='binary:logistic', use_label_encoder=False, eval_metric='error', seed=44)\n",
    "xgb_pipe = make_pipeline(StandardScaler(), xgb)\n",
    "\n",
    "param_grid =   {\n",
    "    'xgbclassifier__n_estimators': [5, 10, 50, 100, 150, 200],              \n",
    "    'xgbclassifier__learning_rate': [0.001, 0.01, 0.1, 0.5],                \n",
    "    'xgbclassifier__max_depth': range(3, 16, 2),                            \n",
    "    'xgbclassifier__min_child_weight': [0.01, 0.05, 0.25, 0.75],            \n",
    "    'xgbclassifier__subsample': [i / 10.0 for i in range(1, 3)],            \n",
    "    'xgbclassifier__colsample_bytree': [i / 10.0 for i in range(1, 3)],     \n",
    "    'xgbclassifier__gamma': [i / 10.0 for i in range(0, 5)]                 \n",
    "}\n",
    "\n",
    "clf = GridSearchCV(xgb_pipe, param_grid, cv=inner_cv)\n",
    "accuracies = cross_val_score(clf, X, y.values.ravel(), cv=outer_cv)\n",
    "\n",
    "xgb_score = accuracies.mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "289d81d9",
   "metadata": {},
   "source": [
    "### Model selection\n",
    "\n",
    "**Logistic regression** was found to have the highest accuracy on this dataset. SVM also performs reasonably well. \n",
    "\n",
    "XGBoost performed surprisingly poorly, which suggests that such non-parametric techniques (e.g. Decision Tree, of which XGBoost is an ensemble) are not effective with this specific dataset. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "id": "920f2f46",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Collate data\n",
    "models = ['Logistic Regression', 'KNN', 'SVM', 'Decision Tree', 'XGBoost']\n",
    "accuracies = np.array([lr_score, neigh_score, svc_score, dt_score, xgb_score])\n",
    "accuracy = pd.DataFrame(data=accuracies, columns=['Accuracy Score'], index=models).sort_values(by='Accuracy Score')\n",
    "\n",
    "plt.figure(figsize=(8, 6))\n",
    "plt.title('Model Performance: Accuracy Score', fontweight='bold')\n",
    "ax = sns.barplot(x=accuracy.index, y='Accuracy Score', data=accuracy)\n",
    "for p in ax.patches:\n",
    "    ax.annotate(format(p.get_height(), '.2f'), \n",
    "                   (p.get_x() + p.get_width() / 2., p.get_height()), \n",
    "                   ha = 'center', va = 'center', \n",
    "                   xytext = (0, 9), \n",
    "                   textcoords = 'offset points')\n",
    "\n",
    "ax.set_ylim(0, 1);"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2d54ce9d",
   "metadata": {},
   "source": [
    "### Logistic Regression: Performance statistics\n",
    "\n",
    "Logistic regression achieves the following headline metrics:\n",
    "\n",
    "- Accuracy: **72%**\n",
    "- Specificity (TNR): **62%**\n",
    "- Sensitivity (TPR): **80%**\n",
    "\n",
    "There is however a trade-off between specificity and sensitivity that can be mapped for logistic regression: for example, sensitivity can be increased at the expense of specificity (and overall accuracy). This is because logistic regression, as a discriminative model, predicts probabilities, and then classifies based on probabilities compared to some cut-off value (by default, 0.5).\n",
    "\n",
    "In this specific context, it would seem to be desirable to prioritise sensitivity to ensure those in ARD are correctly identified (by reducing the cut-off value from 0.5). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "id": "bf4d99e1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot confusion matrix and summary statistics\n",
    "\n",
    "conf = confusion_matrix(y, lr_pred)\n",
    "acc = (conf[0][0] + conf[1][1]) / conf.sum() \n",
    "spec = conf[0][0] / conf[0].sum()\n",
    "sens = conf[1][1] / conf[1].sum()\n",
    "\n",
    "fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(14,5))\n",
    "fig.suptitle('Logistic Regression: Summary Statistics', fontweight='bold')\n",
    "\n",
    "plot_conf_matrix(ax1, conf)\n",
    "\n",
    "plot_bar_chart(ax2, xlabels=['Accuracy', 'Specificity (TNR)', 'Sensitivity (TPR)'], metrics=np.array([acc, spec, sens]), ylabel='Metrics')\n",
    "ax2.set_ylim(0, 1);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "id": "770c1667",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Specificity vs. sensitivity\n",
    "\n",
    "plot_spec_sens(lr_pred_prob[:,1], spec, sens, 'Logistic Regression: Specificity/Sensitivity Trade-off')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6c174ee9",
   "metadata": {},
   "source": [
    "<a id='conclusions'></a>\n",
    "## Conclusions & Next Steps"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cded45d7",
   "metadata": {},
   "source": [
    "It is possible to predict whether a caller to NHS 111 is in ARD with a moderate degree of accuracy. Given the relatively limited data available, this can be seen as a positive result.\n",
    "\n",
    "To improve model performance further, the dataset should be expanded in terms of number of calls (from the current 74), so that more features can be added to the model without suffering from the curse of dimensionality. These features could be:\n",
    "\n",
    "- Additional feature engineering on the current data (e.g. incorporating operator metrics) \n",
    "- Information on the call content itself (e.g. a transcript or presence of certain key words)\n",
    "- Phonetic analysis on the call using software similar to Praat (which has Python bindings through the Parselmouth library). Praat, for example, has been used to count the number of breaths in a recording. There might be issues here, however, with audio quality over the phone.  \n",
    "\n",
    "Another open question is the extent to which only some initial percentage of the call can be used successfully for classification, since in production this model would ideally be used during a call - and not afterwards. "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:root] *",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}