2018/day9_Breast_Cancer_Neural_Network_Classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Predicting Breast Cancer \n",
    "## Part 4 : Neural Network Classification\n",
    "### 1. Load Modules"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import time\n",
    "import sys, os\n",
    "import pandas as pd\n",
    "import helper_functions as hf\n",
    "from helper_functions import training_set_scaled, class_set_scaled\n",
    "from helper_functions import test_set_scaled, test_class_set_scaled\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.model_selection import KFold, cross_val_score \n",
    "from sklearn.model_selection import KFold, GridSearchCV\n",
    "from sklearn.metrics import roc_curve \n",
    "from sklearn.metrics import auc \n",
    "from sklearn.externals import joblib\n",
    "\n",
    "pd.set_option('display.max_columns', 500) \n",
    "# Included to show all the columns \n",
    "# since it is a fairly large data set\n",
    "\n",
    "plt.style.use('ggplot') # Using ggplot2 style visuals "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Loading Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "breast_cancer = hf.breast_cancer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(569, 31)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# dimensions of our data frame\n",
    "breast_cancer.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "diagnosis                    int64\n",
       "radius_mean                float64\n",
       "texture_mean               float64\n",
       "perimeter_mean             float64\n",
       "area_mean                  float64\n",
       "smoothness_mean            float64\n",
       "compactness_mean           float64\n",
       "concavity_mean             float64\n",
       "concave_points_mean        float64\n",
       "symmetry_mean              float64\n",
       "fractal_dimension_mean     float64\n",
       "radius_se                  float64\n",
       "texture_se                 float64\n",
       "perimeter_se               float64\n",
       "area_se                    float64\n",
       "smoothness_se              float64\n",
       "compactness_se             float64\n",
       "concavity_se               float64\n",
       "concave_points_se          float64\n",
       "symmetry_se                float64\n",
       "fractal_dimension_se       float64\n",
       "radius_worst               float64\n",
       "texture_worst              float64\n",
       "perimeter_worst            float64\n",
       "area_worst                 float64\n",
       "smoothness_worst           float64\n",
       "compactness_worst          float64\n",
       "concavity_worst            float64\n",
       "concave_points_worst       float64\n",
       "symmetry_worst             float64\n",
       "fractal_dimension_worst    float64\n",
       "dtype: object"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# data types of our columns \n",
    "breast_cancer.dtypes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Fitting model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "fit_nn = MLPClassifier(solver='lbfgs', \n",
    "                       hidden_layer_sizes = (12, ),\n",
    "                       activation='tanh',\n",
    "                       learning_rate_init=0.05, \n",
    "                       random_state=42)\n",
    "\n",
    "# Train model on training set\n",
    "fit_nn.fit(training_set_scaled, class_set_scaled['diagnosis'])\n",
    "\n",
    "predictions_nn = fit_nn.predict(test_set_scaled)\n",
    "\n",
    "accuracy_nn = fit_nn.score(test_set_scaled, \n",
    "                           test_class_set_scaled['diagnosis'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Here we calculate the test error rate!\n",
    "test_error_rate_nn = 1 - accuracy_nn\n",
    "\n",
    "# ROC Curve stuff\n",
    "fpr3, tpr3, _ = roc_curve(predictions_nn, test_class_set_scaled)\n",
    "\n",
    "auc_nn = auc(fpr3, tpr3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "    ##################################\n",
      "    ##         FITTING MLP          ##\n",
      "    ##################################\n",
      "    \n",
      "MLPClassifier(activation='tanh', alpha=0.0001, batch_size='auto', beta_1=0.9,\n",
      "       beta_2=0.999, early_stopping=False, epsilon=1e-08,\n",
      "       hidden_layer_sizes=(12,), learning_rate='constant',\n",
      "       learning_rate_init=0.05, max_iter=200, momentum=0.9,\n",
      "       nesterovs_momentum=True, power_t=0.5, random_state=42, shuffle=True,\n",
      "       solver='lbfgs', tol=0.0001, validation_fraction=0.1, verbose=False,\n",
      "       warm_start=False)\n",
      "\n",
      "    ############################################\n",
      "    ##       HYPERPARAMETER OPTIMIZATION      ##\n",
      "    ############################################\n",
      "    \n",
      "{'activation': 'tanh', 'hidden_layer_sizes': 12, 'learning_rate_init': 0.05}\n",
      "27.549711227416992\n",
      "\n",
      "    ################################\n",
      "    ##      CROSS VALIDATION      ##\n",
      "    ################################\n",
      "    \n",
      "Accuracy:  0.967 (+/-  0.011)\n",
      "\n",
      "    ###############################\n",
      "    ##   TEST SET CALCULATIONS   ##\n",
      "    ###############################\n",
      "    \n",
      "Actual Values      0   1\n",
      "Predicted Values        \n",
      "0                 68   1\n",
      "1                  3  42\n",
      "Here is our mean accuracy on the test set:\n",
      "  0.965\n",
      "The test error rate for our model is:\n",
      "  0.035\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "if __name__ == '__main__':\n",
    "    print('''\n",
    "    ##################################\n",
    "    ##         FITTING MLP          ##\n",
    "    ##################################\n",
    "    ''')\n",
    "    print(fit_nn)\n",
    "    print('''\n",
    "    ############################################\n",
    "    ##       HYPERPARAMETER OPTIMIZATION      ##\n",
    "    ############################################\n",
    "    ''')\n",
    "    \n",
    "    start = time.time()\n",
    "    gs = GridSearchCV(fit_nn, cv = 10,\n",
    "                      param_grid={\n",
    "                          'learning_rate_init': [0.05, 0.01, 0.005, 0.001],\n",
    "                          'hidden_layer_sizes': [4, 8, 12],\n",
    "                          'activation': [\"relu\", \"identity\", \"tanh\", \"logistic\"]})\n",
    "    gs.fit(training_set_scaled, class_set_scaled['diagnosis'])\n",
    "    print(gs.best_params_)\n",
    "    end = time.time()\n",
    "    print(end - start)\n",
    "    print('''\n",
    "    ################################\n",
    "    ##      CROSS VALIDATION      ##\n",
    "    ################################\n",
    "    ''')\n",
    "    \n",
    "    test_thing = hf.cross_val_metrics(fit_nn, training_set_scaled, \n",
    "                                      class_set_scaled['diagnosis'], \n",
    "                                      print_results = True)\n",
    "    \n",
    "    print('''\n",
    "    ###############################\n",
    "    ##   TEST SET CALCULATIONS   ##\n",
    "    ###############################\n",
    "    ''')\n",
    "    print(pd.crosstab(predictions_nn, \n",
    "                      test_class_set_scaled['diagnosis'], \n",
    "                      rownames=['Predicted Values'], \n",
    "                      colnames=['Actual Values']))\n",
    "\n",
    "    print(\"Here is our mean accuracy on the test set:\\n {0: .3f}\"\n",
    "          .format(accuracy_nn))\n",
    "\n",
    "    print(\"The test error rate for our model is:\\n {0: .3f}\"\n",
    "          .format(test_error_rate_nn))\n",
    "    \n",
    "    # ROC Curve\n",
    "    hf.plot_roc_curve(fpr3, tpr3, auc_nn, 'nn')\n",
    "    \n",
    "    # Zoomed in ROC Curve\n",
    "    hf.plot_roc_curve(fpr3, tpr3, auc_nn, 'nn',  \n",
    "                      (-0.001, 0.2), (0.7, 1.05))\n",
    "else:\n",
    "    def return_nn():\n",
    "        '''\n",
    "        Function to output values created in script \n",
    "        '''\n",
    "        return fpr3, tpr3, auc_nn, predictions_nn, test_error_rate_nn\n",
    "\n",
    "    # Keep Cross validation metrics \n",
    "    mean_cv_nn, std_error_nn = hf.cross_val_metrics(fit_nn, \n",
    "                                                    training_set_scaled, \n",
    "                                                    class_set_scaled['diagnosis'], \n",
    "                                                    print_results = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}