{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img align=\"center\" src=\"http://sydney.edu.au/images/content/about/logo-mono.jpg\">\n",
    "<h1 align=\"center\" style=\"margin-top:10px\">Machine Learning Using Python (MEAFA Workshop)</h1>\n",
    "<h2 align=\"center\" style=\"margin-top:10px\">Lesson 10: Neural Networks (Classification)</h2>\n",
    "<br>\n",
    " \n",
    "<a href=\"#Work-Analytics-Data\">Work Analytics Data</a> <br>\n",
    "<a href=\"#Linear-Support-Vector-Classifier\">Neural Networks</a> <br>\n",
    "<a href=\"#Model-Evaluation\">Model Evaluation</a> <br>\n",
    "\n",
    "This notebook relies on the following libraries and settings."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Packages\n",
    "import numpy as np\n",
    "from scipy import stats\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Plot settings\n",
    "sns.set_context('notebook') \n",
    "sns.set_style('ticks') \n",
    "colours = ['#1F77B4', '#FF7F0E', '#2CA02C', '#DB2728', '#9467BD', '#8C564B', '#E377C2','#7F7F7F', '#BCBD22', '#17BECF']\n",
    "crayon = ['#4E79A7','#F28E2C','#E15759','#76B7B2','#59A14F', '#EDC949','#AF7AA1','#FF9DA7','#9C755F','#BAB0AB']\n",
    "sns.set_palette(colours)\n",
    "%matplotlib inline\n",
    "plt.rcParams['figure.figsize'] = (9, 6)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import GridSearchCV, RandomizedSearchCV\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix, roc_auc_score\n",
    "from sklearn.metrics import precision_score, average_precision_score, log_loss\n",
    "\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Work Analytics Data\n",
    "\n",
    "We use the [Human Resources Analytics](https://www.kaggle.com/ludobenistant/hr-analytics) data available from [Kaggle Datasets](https://www.kaggle.com/datasets).\n",
    "\n",
    "** Business objective: ** To predict which employees will leave the company."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>satisfaction_level</th>\n",
       "      <th>last_evaluation</th>\n",
       "      <th>number_project</th>\n",
       "      <th>average_montly_hours</th>\n",
       "      <th>time_spend_company</th>\n",
       "      <th>Work_accident</th>\n",
       "      <th>left</th>\n",
       "      <th>promotion_last_5years</th>\n",
       "      <th>role</th>\n",
       "      <th>salary</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.38</td>\n",
       "      <td>0.53</td>\n",
       "      <td>2</td>\n",
       "      <td>157</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>sales</td>\n",
       "      <td>low</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.80</td>\n",
       "      <td>0.86</td>\n",
       "      <td>5</td>\n",
       "      <td>262</td>\n",
       "      <td>6</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>sales</td>\n",
       "      <td>medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.11</td>\n",
       "      <td>0.88</td>\n",
       "      <td>7</td>\n",
       "      <td>272</td>\n",
       "      <td>4</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>sales</td>\n",
       "      <td>medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.72</td>\n",
       "      <td>0.87</td>\n",
       "      <td>5</td>\n",
       "      <td>223</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>sales</td>\n",
       "      <td>low</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.37</td>\n",
       "      <td>0.52</td>\n",
       "      <td>2</td>\n",
       "      <td>159</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>sales</td>\n",
       "      <td>low</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   satisfaction_level  last_evaluation  number_project  average_montly_hours  \\\n",
       "0                0.38             0.53               2                   157   \n",
       "1                0.80             0.86               5                   262   \n",
       "2                0.11             0.88               7                   272   \n",
       "3                0.72             0.87               5                   223   \n",
       "4                0.37             0.52               2                   159   \n",
       "\n",
       "   time_spend_company  Work_accident  left  promotion_last_5years   role  \\\n",
       "0                   3              0     1                      0  sales   \n",
       "1                   6              0     1                      0  sales   \n",
       "2                   4              0     1                      0  sales   \n",
       "3                   5              0     1                      0  sales   \n",
       "4                   3              0     1                      0  sales   \n",
       "\n",
       "   salary  \n",
       "0     low  \n",
       "1  medium  \n",
       "2  medium  \n",
       "3     low  \n",
       "4     low  "
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data = pd.read_csv('Datasets/HR.csv')\n",
    "data = data.rename(columns = {'sales' : 'role'})\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "response='left'\n",
    "predictors = list(data.columns.values)\n",
    "predictors.remove(response)\n",
    "\n",
    "index_train, index_test  = train_test_split(np.array(data.index), stratify=data[response], train_size=0.2, random_state=5)\n",
    "\n",
    "train = data.loc[index_train,].copy()\n",
    "test =  data.loc[index_test,:].copy()\n",
    "\n",
    "y_train = train[response]\n",
    "y_test = test[response]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Before estimating the models, we need to convert the categorical variables into binary variables. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dummies = pd.get_dummies(data['role'], drop_first=True)\n",
    "data = data.join(dummies)\n",
    "data = data.drop('role', axis= 1)\n",
    "\n",
    "dummies = pd.get_dummies(data['salary'],  prefix = 'salary', drop_first=True)\n",
    "data = data.join(dummies)\n",
    "data = data.drop('salary', axis= 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We then update our list of predictors accordingly and construct the train and test design matrices. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2999, 18)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = data.loc[index_train,].copy()\n",
    "test =  data.loc[index_test,:].copy()\n",
    "\n",
    "predictors = list(train.columns.values)\n",
    "predictors.remove(response)\n",
    "\n",
    "X_train = data.loc[index_train, predictors].copy()\n",
    "\n",
    "scaler = StandardScaler()\n",
    "scaler.fit(X_train)\n",
    "\n",
    "X_train = scaler.transform(X_train)\n",
    "X_test = scaler.transform(test[predictors].values)\n",
    "\n",
    "X_train.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Neural Networks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Activation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Two Hidden Layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x22f1e4cb898>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mlp = Sequential()\n",
    "mlp.add(Dense(128, input_dim=X_train.shape[1], init='uniform', activation='relu'))\n",
    "mlp.add(Dense(64, init='uniform', activation='relu'))\n",
    "mlp.add(Dense(1, activation='sigmoid'))\n",
    "mlp.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "mlp.fit(X_train, y_train, epochs=500, verbose=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dropout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x22f1ed35e80>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(Dense(128, input_dim=X_train.shape[1], init='uniform', activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(64, init='uniform', activation='relu'))\n",
    "model.add(Dense(1, activation='sigmoid'))\n",
    "model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])\n",
    "model.fit(X_train, y_train, epochs=500, verbose=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Evaluation\n",
    "\n",
    "We estimate a logistic regression as a benchmark."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression(C=100000.0, class_weight=None, dual=False,\n",
       "          fit_intercept=True, intercept_scaling=1, max_iter=100,\n",
       "          multi_class='ovr', n_jobs=1, penalty='l2', random_state=None,\n",
       "          solver='liblinear', tol=0.0001, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "logit = LogisticRegression(C=1e5)\n",
    "logit.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Error rate</th>\n",
       "      <th>Sensitivity</th>\n",
       "      <th>Specificity</th>\n",
       "      <th>AUC</th>\n",
       "      <th>Precision</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Logistic regression</th>\n",
       "      <td>0.225</td>\n",
       "      <td>0.279</td>\n",
       "      <td>0.930</td>\n",
       "      <td>0.822</td>\n",
       "      <td>0.556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Neural Net</th>\n",
       "      <td>0.039</td>\n",
       "      <td>0.922</td>\n",
       "      <td>0.973</td>\n",
       "      <td>0.972</td>\n",
       "      <td>0.915</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Neural Net (dropout)</th>\n",
       "      <td>0.034</td>\n",
       "      <td>0.913</td>\n",
       "      <td>0.982</td>\n",
       "      <td>0.974</td>\n",
       "      <td>0.941</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      Error rate  Sensitivity  Specificity    AUC  Precision\n",
       "Logistic regression        0.225        0.279        0.930  0.822      0.556\n",
       "Neural Net                 0.039        0.922        0.973  0.972      0.915\n",
       "Neural Net (dropout)       0.034        0.913        0.982  0.974      0.941"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "columns=['Error rate', 'Sensitivity', 'Specificity', 'AUC', 'Precision']\n",
    "rows=['Logistic regression', 'Neural Net', 'Neural Net (dropout)']\n",
    "results=pd.DataFrame(0.0, columns=columns, index=rows) \n",
    "\n",
    "methods=[logit, mlp, model]\n",
    "\n",
    "for i, method in enumerate(methods):\n",
    "    \n",
    "    y_pred = method.predict(X_test)\n",
    "    if method != logit:\n",
    "        y_prob = y_pred\n",
    "        y_pred = (y_pred>0.5).astype(int)\n",
    "    else:\n",
    "        y_prob = method.predict_proba(X_test)[:,1]\n",
    "        \n",
    "\n",
    "\n",
    "    confusion  = confusion_matrix(y_test, y_pred)\n",
    "    results.iloc[i,0]=  1 - accuracy_score(y_test, y_pred)\n",
    "    results.iloc[i,1]=  confusion[1,1]/np.sum(confusion[1,:])\n",
    "    results.iloc[i,2]=  confusion[0,0]/np.sum(confusion[0,:])\n",
    "    results.iloc[i,4]=  precision_score(y_test, y_pred)\n",
    "    results.iloc[i,3] = roc_auc_score(y_test, y_prob)\n",
    "\n",
    "results.round(3)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}