neural network analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from sklearn.datasets import make_classification\n",
    "from sklearn.decomposition import PCA\n",
    "from imblearn.over_sampling import SMOTE"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "data=pd.read_csv('상장비상장(2008_2013).csv',header=None,skiprows=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "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>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>28</th>\n",
       "      <th>29</th>\n",
       "      <th>30</th>\n",
       "      <th>31</th>\n",
       "      <th>32</th>\n",
       "      <th>33</th>\n",
       "      <th>34</th>\n",
       "      <th>35</th>\n",
       "      <th>36</th>\n",
       "      <th>37</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>2008.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-11.65</td>\n",
       "      <td>20.34</td>\n",
       "      <td>-30.27</td>\n",
       "      <td>-21.90</td>\n",
       "      <td>8.00</td>\n",
       "      <td>139.78</td>\n",
       "      <td>10.45</td>\n",
       "      <td>56.11</td>\n",
       "      <td>0.95</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2008.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-14.83</td>\n",
       "      <td>73.84</td>\n",
       "      <td>-97.78</td>\n",
       "      <td>-21.95</td>\n",
       "      <td>1.58</td>\n",
       "      <td>9.40</td>\n",
       "      <td>1.91</td>\n",
       "      <td>33.43</td>\n",
       "      <td>1.06</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2008.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-4.54</td>\n",
       "      <td>76.52</td>\n",
       "      <td>7.41</td>\n",
       "      <td>1.27</td>\n",
       "      <td>7.95</td>\n",
       "      <td>71.48</td>\n",
       "      <td>2.81</td>\n",
       "      <td>27.10</td>\n",
       "      <td>5.53</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2008.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>-1.66</td>\n",
       "      <td>-6.21</td>\n",
       "      <td>-1.56</td>\n",
       "      <td>5.31</td>\n",
       "      <td>9.83</td>\n",
       "      <td>-481.27</td>\n",
       "      <td>-31.27</td>\n",
       "      <td>9.51</td>\n",
       "      <td>2.95</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2008.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>6.50</td>\n",
       "      <td>57.73</td>\n",
       "      <td>93.40</td>\n",
       "      <td>161.09</td>\n",
       "      <td>26.41</td>\n",
       "      <td>380.09</td>\n",
       "      <td>10.21</td>\n",
       "      <td>36.98</td>\n",
       "      <td>6.16</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 38 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   0       1    2    3    4    5    6    7    8    9   ...     28     29  \\\n",
       "0   0  2008.0  0.0  1.0  1.0  1.0  0.0  0.0  0.0  0.0  ... -11.65  20.34   \n",
       "1   1  2008.0  0.0  1.0  1.0  1.0  0.0  0.0  0.0  0.0  ... -14.83  73.84   \n",
       "2   2  2008.0  1.0  1.0  0.0  1.0  0.0  0.0  0.0  0.0  ...  -4.54  76.52   \n",
       "3   3  2008.0  0.0  1.0  0.0  1.0  0.0  0.0  0.0  0.0  ...  -1.66  -6.21   \n",
       "4   4  2008.0  0.0  1.0  0.0  1.0  0.0  0.0  0.0  0.0  ...   6.50  57.73   \n",
       "\n",
       "      30      31     32      33     34     35    36   37  \n",
       "0 -30.27  -21.90   8.00  139.78  10.45  56.11  0.95  1.0  \n",
       "1 -97.78  -21.95   1.58    9.40   1.91  33.43  1.06  1.0  \n",
       "2   7.41    1.27   7.95   71.48   2.81  27.10  5.53  1.0  \n",
       "3  -1.56    5.31   9.83 -481.27 -31.27   9.51  2.95  1.0  \n",
       "4  93.40  161.09  26.41  380.09  10.21  36.98  6.16  1.0  \n",
       "\n",
       "[5 rows x 38 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6246, 38)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "input = data.iloc[:,2:37].values\n",
    "output = data.iloc[:,[37]].values "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# We are controlling the seed to compare different models\n",
    "seed = 42\n",
    "tf.set_random_seed(seed)\n",
    "np.random.seed(seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For the training and validation sets we used 2008 to 2012 data as training sets and 2013 data as validation sets.\n",
    "# The reason for not doing it random is because we are interested in how the data we have can predict the future\n",
    "train_indices = []\n",
    "test_indices = []\n",
    "\n",
    "for index,element in data.iterrows():\n",
    "    if element[1] == 2013:\n",
    "        test_indices.append(index)\n",
    "    else:\n",
    "        train_indices.append(index)\n",
    "        \n",
    "train_indices=np.array(train_indices)\n",
    "train_indices=np.int_(train_indices)\n",
    "test_indices=np.array(test_indices)\n",
    "test_indices=np.int_(test_indices)\n",
    "\n",
    "input_train = input[train_indices]\n",
    "input_test = input[test_indices]\n",
    "output_train = output[train_indices]\n",
    "output_test = output[test_indices] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "scaler = StandardScaler()\n",
    "scaler.fit(input_train)\n",
    "input_train = scaler.transform(input_train)\n",
    "input_test = scaler.transform(input_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Before OverSampling, the shape of input_train: (5205, 35)\n",
      "Before OverSampling, the shape of output_train: (5205, 1) \n",
      "\n",
      "Before OverSampling, counts of label '1': [222]\n",
      "Before OverSampling, counts of label '0': [4983]\n"
     ]
    }
   ],
   "source": [
    "print('Before OverSampling, the shape of input_train: {}'.format(input_train.shape))\n",
    "print('Before OverSampling, the shape of output_train: {} \\n'.format(output_train.shape))\n",
    "\n",
    "print(\"Before OverSampling, counts of label '1': {}\".format(sum(output_train==1)))\n",
    "print(\"Before OverSampling, counts of label '0': {}\".format(sum(output_train==0)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\youngjae.kim\\AppData\\Local\\Continuum\\anaconda3\\lib\\site-packages\\sklearn\\utils\\validation.py:578: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    }
   ],
   "source": [
    "# In order to solve the data imbalance problem we used the oversampling method.\n",
    "sm = SMOTE(ratio='auto', kind='regular')\n",
    "input_train, output_train = sm.fit_sample(input_train,list(output_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# The following are some under sampling methods\n",
    "\n",
    "#from imblearn.under_sampling import TomekLinks\n",
    "#tl = TomekLinks(return_indices=True, ratio='majority')\n",
    "#input_train, output_train, id_tl = tl.fit_sample(input_train, output_train)\n",
    "\n",
    "#from imblearn.under_sampling import ClusterCentroids\n",
    "#cc = ClusterCentroids(ratio={0: 300})\n",
    "#input_train, output_train = cc.fit_sample(input_train, output_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "After OverSampling, the shape of input_train: (9966, 35)\n",
      "After OverSampling, the shape of output_train: (9966,) \n",
      "\n",
      "After OverSampling, counts of label '1': 4983\n",
      "After OverSampling, counts of label '0': 4983\n"
     ]
    }
   ],
   "source": [
    "print('After OverSampling, the shape of input_train: {}'.format(input_train.shape))\n",
    "print('After OverSampling, the shape of output_train: {} \\n'.format(output_train.shape))\n",
    "\n",
    "print(\"After OverSampling, counts of label '1': {}\".format(sum(output_train==1)))\n",
    "print(\"After OverSampling, counts of label '0': {}\".format(sum(output_train==0)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "epoch_size= 1500\n",
    "batch_size= 2048\n",
    "hidden_size1=2\n",
    "hidden_size2=8\n",
    "result=1\n",
    "n=input_train[0].size\n",
    "m=output_train.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "output_train = np.array(output_train).reshape(m,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = tf.placeholder(shape=[None, n], dtype=tf.float32)\n",
    "Y = tf.placeholder(shape=[None, 1], dtype=tf.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0916 05:40:28.599570 10428 lazy_loader.py:50] \n",
      "The TensorFlow contrib module will not be included in TensorFlow 2.0.\n",
      "For more information, please see:\n",
      "  * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md\n",
      "  * https://github.com/tensorflow/addons\n",
      "  * https://github.com/tensorflow/io (for I/O related ops)\n",
      "If you depend on functionality not listed there, please file an issue.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "initializer = tf.contrib.layers.variance_scaling_initializer()\n",
    "\n",
    "w1 = tf.Variable(initializer(shape=[n, hidden_size1]))\n",
    "b1 = tf.Variable(initializer(shape=[hidden_size1]))\n",
    "#w2 = tf.Variable(initializer(shape=[hidden_size1,result]))\n",
    "#b2 = tf.Variable(initializer(shape=[result]))\n",
    "w2 = tf.Variable(initializer(shape=[hidden_size1,hidden_size2]))\n",
    "b2 = tf.Variable(initializer(shape=[hidden_size2]))\n",
    "w3 = tf.Variable(initializer(shape=[hidden_size2, result]))\n",
    "b3 = tf.Variable(initializer(shape=[result]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "hidden1 = tf.nn.relu(tf.add(tf.matmul(X, w1), b1))\n",
    "hidden2 = tf.nn.relu(tf.add(tf.matmul(hidden1, w2), b2))\n",
    "#final = tf.add(tf.matmul(hidden1, w2), b2)\n",
    "final = tf.add(tf.matmul(hidden2, w3), b3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#since it is more important to find companies that has dept we set an extra weight on these companies\n",
    "loss = tf.reduce_mean(tf.nn.weighted_cross_entropy_with_logits(logits=final,labels=Y,pos_weight=1.4))\n",
    "global_step = tf.Variable(0, trainable=False)\n",
    "starter_learning_rate = 0.01\n",
    "learning_rate = tf.train.exponential_decay(starter_learning_rate, global_step,epoch_size, 0.99, staircase=True)\n",
    "train_step = (\n",
    "    tf.train.AdamOptimizer(learning_rate)\n",
    "    .minimize(loss, global_step=global_step)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "prediction = tf.round(tf.sigmoid(final))\n",
    "correct = tf.cast(tf.equal(prediction, Y), dtype=tf.float32)\n",
    "accuracy = tf.reduce_mean(correct)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "sess=tf.Session()\n",
    "init = tf.global_variables_initializer()\n",
    "sess.run(init)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Generation: 100. Train Loss = 0.6200199. Test Loss = 0.94749624. Train Accuracy = 0.7373047. Test Accuracy = 0.73775214\n",
      "Generation: 200. Train Loss = 0.5867667. Test Loss = 1.1366392. Train Accuracy = 0.7504883. Test Accuracy = 0.7223823\n",
      "Generation: 300. Train Loss = 0.5413716. Test Loss = 1.602864. Train Accuracy = 0.77783203. Test Accuracy = 0.70989436\n",
      "Generation: 400. Train Loss = 0.5258043. Test Loss = 2.286514. Train Accuracy = 0.77441406. Test Accuracy = 0.7060519\n",
      "Generation: 500. Train Loss = 0.5062187. Test Loss = 2.814145. Train Accuracy = 0.7944336. Test Accuracy = 0.71757925\n",
      "Generation: 600. Train Loss = 0.50036657. Test Loss = 3.2006457. Train Accuracy = 0.7919922. Test Accuracy = 0.7012488\n",
      "Generation: 700. Train Loss = 0.49122703. Test Loss = 3.8137467. Train Accuracy = 0.7895508. Test Accuracy = 0.685879\n",
      "Generation: 800. Train Loss = 0.52316594. Test Loss = 4.304284. Train Accuracy = 0.7788086. Test Accuracy = 0.6820365\n",
      "Generation: 900. Train Loss = 0.49820358. Test Loss = 4.7798448. Train Accuracy = 0.7988281. Test Accuracy = 0.67819405\n",
      "Generation: 1000. Train Loss = 0.48848784. Test Loss = 5.3227057. Train Accuracy = 0.7919922. Test Accuracy = 0.685879\n",
      "Generation: 1100. Train Loss = 0.5067009. Test Loss = 5.560393. Train Accuracy = 0.79296875. Test Accuracy = 0.6753122\n",
      "Generation: 1200. Train Loss = 0.49606055. Test Loss = 5.419055. Train Accuracy = 0.79003906. Test Accuracy = 0.6810759\n",
      "Generation: 1300. Train Loss = 0.4886331. Test Loss = 4.9615755. Train Accuracy = 0.80126953. Test Accuracy = 0.685879\n",
      "Generation: 1400. Train Loss = 0.49130785. Test Loss = 4.407011. Train Accuracy = 0.8017578. Test Accuracy = 0.69068205\n",
      "Generation: 1500. Train Loss = 0.48591587. Test Loss = 4.4260416. Train Accuracy = 0.7998047. Test Accuracy = 0.673391\n",
      "Final Train Accuracy: 0.7998047\n",
      "Final Test Accuracy: 0.673391\n"
     ]
    }
   ],
   "source": [
    "train_loss = []\n",
    "test_loss = []\n",
    "train_accu= []\n",
    "test_accu=[]\n",
    "\n",
    "for i in range(epoch_size):\n",
    "    rand_index = np.random.choice(len(input_train), size=batch_size)\n",
    "    rand_x = input_train[rand_index]\n",
    "    rand_y = output_train[rand_index]\n",
    "    \n",
    "    sess.run(train_step, feed_dict={X: rand_x, Y: rand_y})\n",
    "    \n",
    "    loss1 = sess.run(loss, feed_dict={X: rand_x, Y: rand_y})\n",
    "    train_loss.append(loss1)\n",
    "    \n",
    "    accu1 = sess.run(accuracy, feed_dict={X: rand_x, Y: output_train[rand_index]})\n",
    "    train_accu.append(accu1)\n",
    "    \n",
    "    loss2 = sess.run(loss, feed_dict={X: input_test, Y: output_test})\n",
    "    test_loss.append(loss2)\n",
    "    \n",
    "    accu2 = sess.run(accuracy, feed_dict={X: input_test, Y: output_test})\n",
    "    test_accu.append(np.sqrt(accu2))\n",
    "    \n",
    "    if (i+1)%100==0:\n",
    "        print('Generation: ' + str(i+1) + '. Train Loss = ' + str(loss1)+ '. Test Loss = ' + str(loss2) + \n",
    "              '. Train Accuracy = ' + str(accu1) + '. Test Accuracy = ' + str(accu2))\n",
    "        \n",
    "print('Final Train Accuracy: ' + str(accu1))\n",
    "print('Final Test Accuracy: ' + str(accu2))  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "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": [
    "# The plot of the test accuracy goes down since instead of find the optimal accuracy point we are finding a model that can predict\n",
    "# Companies that will fail to cover their dept\n",
    "plt.plot(train_loss, 'k-', label='Train Loss')\n",
    "plt.plot(test_loss, 'r--', label='Test Loss')\n",
    "plt.title('Loss (MSE) per Generation')\n",
    "plt.xlabel('Generation')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = sess.run(prediction, feed_dict={X: input})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import precision_score\n",
    "from sklearn.metrics import recall_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "target_names = ['false', 'true']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[3467 1516]\n",
      " [ 460 4523]]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "      false       0.88      0.70      0.78      4983\n",
      "       true       0.75      0.91      0.82      4983\n",
      "\n",
      "avg / total       0.82      0.80      0.80      9966\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.8017258679510335"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_train = sess.run(prediction, feed_dict={X: input_train})\n",
    "print(confusion_matrix(output_train.flatten(), pred_train))\n",
    "print(classification_report(output_train.flatten(), pred_train, target_names=target_names))\n",
    "precision_score(output_train.flatten(), pred_train, average='micro',pos_label=1)\n",
    "recall_score(output_train.flatten(), pred_train, average='micro',pos_label=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[661 331]\n",
      " [  9  40]]\n",
      "             precision    recall  f1-score   support\n",
      "\n",
      "      false       0.99      0.67      0.80       992\n",
      "       true       0.11      0.82      0.19        49\n",
      "\n",
      "avg / total       0.95      0.67      0.77      1041\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.6733909702209414"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# The most value we are interested in is the value of recall for 'true' So for this cas 0.82 is the critical number\n",
    "# So in this case out of the 49 companies that become bankrupt 40 of them we were able to track down.\n",
    "pred_test = sess.run(prediction, feed_dict={X: input_test})\n",
    "print(confusion_matrix(output_test.flatten(), pred_test))\n",
    "print(classification_report(output_test.flatten(), pred_test, target_names=target_names))\n",
    "precision_score(output_test.flatten(), pred_test, average='micro',pos_label=1)\n",
    "recall_score(output_test.flatten(), pred_test, average='micro',pos_label=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import metrics\n",
    "import matplotlib.pyplot as plt "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_test = sess.run(final, feed_dict={X: input_test})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "fpr, tpr, thresholds = metrics.roc_curve(output_test, pred_test)\n",
    "roc_auc = metrics.auc(fpr, tpr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x24e3bb48128>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "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": [
    "plt.plot(fpr, tpr, label='ROC curve (area = %0.3f)' % roc_auc)\n",
    "plt.plot([0, 1], [0, 1], 'k--')  # random predictions curve\n",
    "plt.xlim([0.0, 1.0])\n",
    "plt.ylim([0.0, 1.0])\n",
    "plt.xlabel('False Positive Rate or (1 - Specifity)')\n",
    "plt.ylabel('True Positive Rate or (Sensitivity)')\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.legend(loc=\"lower right\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "hidden = sess.run(hidden1, feed_dict={X: input_train})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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": [
    "from matplotlib import pyplot as plt\n",
    "\n",
    "scatter_x = hidden[:,0]\n",
    "scatter_y = hidden[:,1]\n",
    "group = output_train.flatten()\n",
    "cdict = {0:'blue',1:'red'}\n",
    "\n",
    "fig, ax = plt.subplots()\n",
    "for g in np.unique(group):\n",
    "    ix = np.where(group == g)\n",
    "    ax.scatter(scatter_x[ix], scatter_y[ix], c = cdict[g], label = g, s = 100)\n",
    "ax.legend()\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}