  Implimenting Tensorflow and Keras on Fashion-MNIST v2.ipynb

{
  "cells": [
    {
      "metadata": {
        "_uuid": "3b2fb25e2ab812829bf74f624b40ed2f664e4636"
      },
      "cell_type": "markdown",
      "source": "## Data Description\nEach image is 28 pixels in height and 28 pixels in width, for a total of 784 pixels in total. Each pixel has a single pixel-value associated with it, indicating the lightness or darkness of that pixel, with higher numbers meaning darker. This pixel-value is an integer between 0 and 255. The training and test data sets have 785 columns. The first column consists of the class labels (see above), and represents the article of clothing. The rest of the columns contain the pixel-values of the associated image.\n### Labels\nEach training and test example is assigned to one of the following labels:\n\n0 T-shirt/top\n\n1 Trouser\n\n2 Pullover\n\n3 Dress\n\n4 Coat\n\n5 Sandal\n\n6 Shirt\n\n7 Sneaker\n\n8 Bag\n\n9 Ankle boot\n\n\n\nEach row is a separate image\nColumn 1 is the class label.\nRemaining columns are pixel numbers (784 total).\nEach value is the darkness of the pixel (1 to 255)"
    },
    {
      "metadata": {
        "_uuid": "6190cb0690947063098cc00f7cccd23bb07643dd"
      },
      "cell_type": "markdown",
      "source": "### Loading useful libraries"
    },
    {
      "metadata": {
        "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
        "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
        "trusted": true
      },
      "cell_type": "code",
      "source": "# Input      data     files     are    available in the \"../input/\" directory.\n\nfrom subprocess import check_output\n\nprint(check_output([\"ls\", \"../input\"]).decode(\"utf8\"))\n\n# Import Pandas for data manipulation using dataframes\n\nimport pandas as pd\n\n#Import Numpy for statistical calculations\n\nimport numpy as np\n\n# Import Warnings \n\nimport warnings\n\nwarnings.filterwarnings('ignore')\n\n# Import matplotlib Library for data visualisation\n\nimport matplotlib.pyplot as plt\n\n#Import train_test_split from scikit library\n\nfrom sklearn.model_selection import train_test_split\n\n# Import Keras\n\nimport keras\n\nfrom keras.models import Sequential\n\nfrom keras.layers import Conv2D,MaxPooling2D,Dense,Flatten,Dropout\n\nfrom keras.optimizers import Adam\n\nfrom keras.callbacks import TensorBoard\n\nnum_classes = 10\n\nepochs = 5\n",
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": "fashion-mnist_test.csv\nfashion-mnist_train.csv\nt10k-images-idx3-ubyte\nt10k-labels-idx1-ubyte\ntrain-images-idx3-ubyte\ntrain-labels-idx1-ubyte\n\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
        "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a",
        "trusted": true
      },
      "cell_type": "code",
      "source": "#Loading test and train data\ntrain_df = pd.read_csv('/kaggle/input/fashion-mnist_train.csv',sep=',')\ntest_df = pd.read_csv('/kaggle/input/fashion-mnist_test.csv', sep = ',')",
      "execution_count": 18,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "baa8329ad78e6fe6d0cb9a81c7d4d4b19c1d0ca5"
      },
      "cell_type": "code",
      "source": "train_df.head()",
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 19,
          "data": {
            "text/plain": "   label  pixel1  pixel2    ...     pixel782  pixel783  pixel784\n0      2       0       0    ...            0         0         0\n1      9       0       0    ...            0         0         0\n2      6       0       0    ...            0         0         0\n3      0       0       0    ...            0         0         0\n4      3       0       0    ...            0         0         0\n\n[5 rows x 785 columns]",
            "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>label</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>pixel10</th>\n      <th>pixel11</th>\n      <th>pixel12</th>\n      <th>pixel13</th>\n      <th>pixel14</th>\n      <th>pixel15</th>\n      <th>pixel16</th>\n      <th>pixel17</th>\n      <th>pixel18</th>\n      <th>pixel19</th>\n      <th>pixel20</th>\n      <th>pixel21</th>\n      <th>pixel22</th>\n      <th>pixel23</th>\n      <th>pixel24</th>\n      <th>pixel25</th>\n      <th>pixel26</th>\n      <th>pixel27</th>\n      <th>pixel28</th>\n      <th>pixel29</th>\n      <th>pixel30</th>\n      <th>pixel31</th>\n      <th>pixel32</th>\n      <th>pixel33</th>\n      <th>pixel34</th>\n      <th>pixel35</th>\n      <th>pixel36</th>\n      <th>pixel37</th>\n      <th>pixel38</th>\n      <th>pixel39</th>\n      <th>...</th>\n      <th>pixel745</th>\n      <th>pixel746</th>\n      <th>pixel747</th>\n      <th>pixel748</th>\n      <th>pixel749</th>\n      <th>pixel750</th>\n      <th>pixel751</th>\n      <th>pixel752</th>\n      <th>pixel753</th>\n      <th>pixel754</th>\n      <th>pixel755</th>\n      <th>pixel756</th>\n      <th>pixel757</th>\n      <th>pixel758</th>\n      <th>pixel759</th>\n      <th>pixel760</th>\n      <th>pixel761</th>\n      <th>pixel762</th>\n      <th>pixel763</th>\n      <th>pixel764</th>\n      <th>pixel765</th>\n      <th>pixel766</th>\n      <th>pixel767</th>\n      <th>pixel768</th>\n      <th>pixel769</th>\n      <th>pixel770</th>\n      <th>pixel771</th>\n      <th>pixel772</th>\n      <th>pixel773</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n      <th>pixel784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>6</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>5</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>105</td>\n      <td>92</td>\n      <td>101</td>\n      <td>107</td>\n      <td>100</td>\n      <td>132</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>4</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>150</td>\n      <td>...</td>\n      <td>211</td>\n      <td>220</td>\n      <td>214</td>\n      <td>74</td>\n      <td>0</td>\n      <td>255</td>\n      <td>222</td>\n      <td>128</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>44</td>\n      <td>12</td>\n      <td>0</td>\n      <td>0</td>\n      <td>40</td>\n      <td>134</td>\n      <td>162</td>\n      <td>191</td>\n      <td>214</td>\n      <td>163</td>\n      <td>146</td>\n      <td>165</td>\n      <td>79</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>30</td>\n      <td>43</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>114</td>\n      <td>183</td>\n      <td>112</td>\n      <td>55</td>\n      <td>23</td>\n      <td>72</td>\n      <td>102</td>\n      <td>165</td>\n      <td>160</td>\n      <td>28</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>24</td>\n      <td>188</td>\n      <td>163</td>\n      <td>93</td>\n      <td>...</td>\n      <td>171</td>\n      <td>249</td>\n      <td>207</td>\n      <td>197</td>\n      <td>202</td>\n      <td>45</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>22</td>\n      <td>21</td>\n      <td>25</td>\n      <td>69</td>\n      <td>52</td>\n      <td>45</td>\n      <td>74</td>\n      <td>39</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>46</td>\n      <td>0</td>\n      <td>21</td>\n      <td>68</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>25</td>\n      <td>187</td>\n      <td>189</td>\n      <td>...</td>\n      <td>230</td>\n      <td>237</td>\n      <td>229</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>68</td>\n      <td>116</td>\n      <td>112</td>\n      <td>136</td>\n      <td>147</td>\n      <td>144</td>\n      <td>121</td>\n      <td>102</td>\n      <td>63</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "ba9a7fdb92a7ec9ea46165d29580e91cec7f0532"
      },
      "cell_type": "code",
      "source": "test_df.head()",
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 20,
          "data": {
            "text/plain": "   label  pixel1  pixel2    ...     pixel782  pixel783  pixel784\n0      0       0       0    ...            0         0         0\n1      1       0       0    ...            0         0         0\n2      2       0       0    ...            0         0         0\n3      2       0       0    ...           56         0         0\n4      3       0       0    ...            0         0         0\n\n[5 rows x 785 columns]",
            "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>label</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>pixel10</th>\n      <th>pixel11</th>\n      <th>pixel12</th>\n      <th>pixel13</th>\n      <th>pixel14</th>\n      <th>pixel15</th>\n      <th>pixel16</th>\n      <th>pixel17</th>\n      <th>pixel18</th>\n      <th>pixel19</th>\n      <th>pixel20</th>\n      <th>pixel21</th>\n      <th>pixel22</th>\n      <th>pixel23</th>\n      <th>pixel24</th>\n      <th>pixel25</th>\n      <th>pixel26</th>\n      <th>pixel27</th>\n      <th>pixel28</th>\n      <th>pixel29</th>\n      <th>pixel30</th>\n      <th>pixel31</th>\n      <th>pixel32</th>\n      <th>pixel33</th>\n      <th>pixel34</th>\n      <th>pixel35</th>\n      <th>pixel36</th>\n      <th>pixel37</th>\n      <th>pixel38</th>\n      <th>pixel39</th>\n      <th>...</th>\n      <th>pixel745</th>\n      <th>pixel746</th>\n      <th>pixel747</th>\n      <th>pixel748</th>\n      <th>pixel749</th>\n      <th>pixel750</th>\n      <th>pixel751</th>\n      <th>pixel752</th>\n      <th>pixel753</th>\n      <th>pixel754</th>\n      <th>pixel755</th>\n      <th>pixel756</th>\n      <th>pixel757</th>\n      <th>pixel758</th>\n      <th>pixel759</th>\n      <th>pixel760</th>\n      <th>pixel761</th>\n      <th>pixel762</th>\n      <th>pixel763</th>\n      <th>pixel764</th>\n      <th>pixel765</th>\n      <th>pixel766</th>\n      <th>pixel767</th>\n      <th>pixel768</th>\n      <th>pixel769</th>\n      <th>pixel770</th>\n      <th>pixel771</th>\n      <th>pixel772</th>\n      <th>pixel773</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n      <th>pixel784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>9</td>\n      <td>8</td>\n      <td>0</td>\n      <td>0</td>\n      <td>34</td>\n      <td>29</td>\n      <td>7</td>\n      <td>0</td>\n      <td>11</td>\n      <td>24</td>\n      <td>0</td>\n      <td>0</td>\n      <td>3</td>\n      <td>3</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>4</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>242</td>\n      <td>245</td>\n      <td>224</td>\n      <td>245</td>\n      <td>234</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>37</td>\n      <td>69</td>\n      <td>94</td>\n      <td>123</td>\n      <td>127</td>\n      <td>138</td>\n      <td>138</td>\n      <td>142</td>\n      <td>145</td>\n      <td>135</td>\n      <td>125</td>\n      <td>103</td>\n      <td>87</td>\n      <td>56</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>209</td>\n      <td>190</td>\n      <td>181</td>\n      <td>150</td>\n      <td>170</td>\n      <td>193</td>\n      <td>180</td>\n      <td>219</td>\n      <td>5</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>24</td>\n      <td>...</td>\n      <td>248</td>\n      <td>238</td>\n      <td>80</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>174</td>\n      <td>233</td>\n      <td>155</td>\n      <td>0</td>\n      <td>65</td>\n      <td>235</td>\n      <td>216</td>\n      <td>34</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>14</td>\n      <td>53</td>\n      <td>99</td>\n      <td>17</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>12</td>\n      <td>94</td>\n      <td>68</td>\n      <td>14</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>38</td>\n      <td>106</td>\n      <td>94</td>\n      <td>89</td>\n      <td>94</td>\n      <td>68</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>87</td>\n      <td>12</td>\n      <td>26</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>46</td>\n      <td>50</td>\n      <td>29</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>63</td>\n      <td>53</td>\n      <td>31</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>161</td>\n      <td>212</td>\n      <td>138</td>\n      <td>150</td>\n      <td>169</td>\n      <td>164</td>\n      <td>176</td>\n      <td>202</td>\n      <td>255</td>\n      <td>183</td>\n      <td>26</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>129</td>\n      <td>221</td>\n      <td>255</td>\n      <td>255</td>\n      <td>215</td>\n      <td>...</td>\n      <td>255</td>\n      <td>255</td>\n      <td>254</td>\n      <td>252</td>\n      <td>255</td>\n      <td>101</td>\n      <td>196</td>\n      <td>254</td>\n      <td>252</td>\n      <td>100</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>168</td>\n      <td>232</td>\n      <td>225</td>\n      <td>0</td>\n      <td>41</td>\n      <td>172</td>\n      <td>147</td>\n      <td>148</td>\n      <td>153</td>\n      <td>155</td>\n      <td>146</td>\n      <td>137</td>\n      <td>141</td>\n      <td>143</td>\n      <td>137</td>\n      <td>137</td>\n      <td>126</td>\n      <td>140</td>\n      <td>0</td>\n      <td>133</td>\n      <td>224</td>\n      <td>222</td>\n      <td>56</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>37</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>17</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>21</td>\n      <td>204</td>\n      <td>235</td>\n      <td>...</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>2</td>\n      <td>69</td>\n      <td>0</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "526bafe18be4e17fbc0fe307006a2205aa1a93e0"
      },
      "cell_type": "markdown",
      "source": "Now let us split the train data into x and y arrays where x represents the image data and y represents the labels.\n\nTo do that we need to convert the dataframes into numpy arrays of float32 type which is the acceptable form for tensorflow and keras.\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "3d45b24549df174152de3e7b32d52098091a0506"
      },
      "cell_type": "code",
      "source": "\ntrain_data = np.array(train_df, dtype = 'float32')\n",
      "execution_count": 21,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "a5609f34959bf650e77a4fdd0f4021919f0a8130"
      },
      "cell_type": "code",
      "source": "#Similarly let us do the same process for test data\n\ntest_data = np.array(test_df, dtype='float32')\n",
      "execution_count": 22,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "853aa4daf322a190818aa51d0441a96e800a51ec"
      },
      "cell_type": "markdown",
      "source": "Now let us slice the train arrays into x and y arrays namely x_train,y_train to store all image data and label data respectively. i.e\n\n    x_train contains all the rows and all columns except the label column and excluding header info .\n\n    y_train contains all the rows and first column and excluding header info .\n\nSimilarly slice the test arrays into x and y arrays namely x_test,y_test to store all image data and label data respectively. i.e\n\n    x_test contains all the rows and all columns except the label column and excluding header info .\n    y_test contains all the rows and first column and excluding header info .\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "fbb97e88df53930ca74286c6e544be94cca1262d"
      },
      "cell_type": "code",
      "source": "train_df.describe()",
      "execution_count": 23,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 23,
          "data": {
            "text/plain": "              label        pixel1     ...           pixel783     pixel784\ncount  60000.000000  60000.000000     ...       60000.000000  60000.00000\nmean       4.500000      0.000900     ...           0.855517      0.07025\nstd        2.872305      0.094689     ...           9.356960      2.12587\nmin        0.000000      0.000000     ...           0.000000      0.00000\n25%        2.000000      0.000000     ...           0.000000      0.00000\n50%        4.500000      0.000000     ...           0.000000      0.00000\n75%        7.000000      0.000000     ...           0.000000      0.00000\nmax        9.000000     16.000000     ...         255.000000    170.00000\n\n[8 rows x 785 columns]",
            "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>label</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>pixel10</th>\n      <th>pixel11</th>\n      <th>pixel12</th>\n      <th>pixel13</th>\n      <th>pixel14</th>\n      <th>pixel15</th>\n      <th>pixel16</th>\n      <th>pixel17</th>\n      <th>pixel18</th>\n      <th>pixel19</th>\n      <th>pixel20</th>\n      <th>pixel21</th>\n      <th>pixel22</th>\n      <th>pixel23</th>\n      <th>pixel24</th>\n      <th>pixel25</th>\n      <th>pixel26</th>\n      <th>pixel27</th>\n      <th>pixel28</th>\n      <th>pixel29</th>\n      <th>pixel30</th>\n      <th>pixel31</th>\n      <th>pixel32</th>\n      <th>pixel33</th>\n      <th>pixel34</th>\n      <th>pixel35</th>\n      <th>pixel36</th>\n      <th>pixel37</th>\n      <th>pixel38</th>\n      <th>pixel39</th>\n      <th>...</th>\n      <th>pixel745</th>\n      <th>pixel746</th>\n      <th>pixel747</th>\n      <th>pixel748</th>\n      <th>pixel749</th>\n      <th>pixel750</th>\n      <th>pixel751</th>\n      <th>pixel752</th>\n      <th>pixel753</th>\n      <th>pixel754</th>\n      <th>pixel755</th>\n      <th>pixel756</th>\n      <th>pixel757</th>\n      <th>pixel758</th>\n      <th>pixel759</th>\n      <th>pixel760</th>\n      <th>pixel761</th>\n      <th>pixel762</th>\n      <th>pixel763</th>\n      <th>pixel764</th>\n      <th>pixel765</th>\n      <th>pixel766</th>\n      <th>pixel767</th>\n      <th>pixel768</th>\n      <th>pixel769</th>\n      <th>pixel770</th>\n      <th>pixel771</th>\n      <th>pixel772</th>\n      <th>pixel773</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n      <th>pixel784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>count</th>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.00000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>...</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.00000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.000000</td>\n      <td>60000.00000</td>\n    </tr>\n    <tr>\n      <th>mean</th>\n      <td>4.500000</td>\n      <td>0.000900</td>\n      <td>0.006150</td>\n      <td>0.035333</td>\n      <td>0.101933</td>\n      <td>0.247967</td>\n      <td>0.411467</td>\n      <td>0.805767</td>\n      <td>2.198283</td>\n      <td>5.682000</td>\n      <td>14.488767</td>\n      <td>32.830583</td>\n      <td>45.731017</td>\n      <td>52.064083</td>\n      <td>50.80245</td>\n      <td>50.459150</td>\n      <td>53.796217</td>\n      <td>51.396233</td>\n      <td>40.432733</td>\n      <td>26.083333</td>\n      <td>10.420583</td>\n      <td>4.079750</td>\n      <td>1.938183</td>\n      <td>1.177650</td>\n      <td>0.857983</td>\n      <td>0.461133</td>\n      <td>0.244300</td>\n      <td>0.092667</td>\n      <td>0.018333</td>\n      <td>0.006400</td>\n      <td>0.020617</td>\n      <td>0.098967</td>\n      <td>0.357200</td>\n      <td>0.767217</td>\n      <td>2.104583</td>\n      <td>7.331483</td>\n      <td>17.472317</td>\n      <td>32.009017</td>\n      <td>52.814583</td>\n      <td>79.349917</td>\n      <td>...</td>\n      <td>92.470233</td>\n      <td>90.523350</td>\n      <td>79.255767</td>\n      <td>62.576500</td>\n      <td>45.753283</td>\n      <td>43.970183</td>\n      <td>47.772400</td>\n      <td>38.841583</td>\n      <td>22.306933</td>\n      <td>9.280517</td>\n      <td>3.955417</td>\n      <td>0.631000</td>\n      <td>0.024633</td>\n      <td>0.218117</td>\n      <td>1.027250</td>\n      <td>4.352033</td>\n      <td>12.568183</td>\n      <td>20.847967</td>\n      <td>21.076467</td>\n      <td>17.152267</td>\n      <td>22.114933</td>\n      <td>30.559350</td>\n      <td>41.743617</td>\n      <td>51.506583</td>\n      <td>52.60855</td>\n      <td>46.479783</td>\n      <td>42.612733</td>\n      <td>45.309733</td>\n      <td>50.162100</td>\n      <td>46.438617</td>\n      <td>34.625400</td>\n      <td>23.300683</td>\n      <td>16.588267</td>\n      <td>17.869433</td>\n      <td>22.814817</td>\n      <td>17.911483</td>\n      <td>8.520633</td>\n      <td>2.753300</td>\n      <td>0.855517</td>\n      <td>0.07025</td>\n    </tr>\n    <tr>\n      <th>std</th>\n      <td>2.872305</td>\n      <td>0.094689</td>\n      <td>0.271011</td>\n      <td>1.222324</td>\n      <td>2.452871</td>\n      <td>4.306912</td>\n      <td>5.836188</td>\n      <td>8.215169</td>\n      <td>14.093378</td>\n      <td>23.819481</td>\n      <td>38.334549</td>\n      <td>57.633860</td>\n      <td>65.964300</td>\n      <td>68.603587</td>\n      <td>67.86885</td>\n      <td>67.370188</td>\n      <td>69.467796</td>\n      <td>67.912739</td>\n      <td>62.639596</td>\n      <td>51.818389</td>\n      <td>32.591095</td>\n      <td>20.118496</td>\n      <td>13.953210</td>\n      <td>10.718718</td>\n      <td>9.256114</td>\n      <td>6.768024</td>\n      <td>4.953103</td>\n      <td>2.850576</td>\n      <td>1.232175</td>\n      <td>0.548966</td>\n      <td>1.184869</td>\n      <td>2.912779</td>\n      <td>6.523878</td>\n      <td>9.702476</td>\n      <td>15.977721</td>\n      <td>30.056073</td>\n      <td>47.953585</td>\n      <td>65.625544</td>\n      <td>82.122567</td>\n      <td>95.340204</td>\n      <td>...</td>\n      <td>100.912013</td>\n      <td>99.731484</td>\n      <td>95.172029</td>\n      <td>88.844198</td>\n      <td>78.434831</td>\n      <td>76.970849</td>\n      <td>83.212136</td>\n      <td>74.477031</td>\n      <td>55.983447</td>\n      <td>37.937984</td>\n      <td>24.953295</td>\n      <td>8.146274</td>\n      <td>0.968474</td>\n      <td>4.444904</td>\n      <td>10.030021</td>\n      <td>20.912278</td>\n      <td>36.920905</td>\n      <td>48.909477</td>\n      <td>48.263137</td>\n      <td>42.173000</td>\n      <td>48.241135</td>\n      <td>54.902062</td>\n      <td>61.963606</td>\n      <td>68.331468</td>\n      <td>69.76041</td>\n      <td>67.476669</td>\n      <td>66.267004</td>\n      <td>66.737520</td>\n      <td>68.705817</td>\n      <td>65.944847</td>\n      <td>57.545242</td>\n      <td>48.854427</td>\n      <td>41.979611</td>\n      <td>43.966032</td>\n      <td>51.830477</td>\n      <td>45.149388</td>\n      <td>29.614859</td>\n      <td>17.397652</td>\n      <td>9.356960</td>\n      <td>2.12587</td>\n    </tr>\n    <tr>\n      <th>min</th>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n    </tr>\n    <tr>\n      <th>25%</th>\n      <td>2.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>...</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n    </tr>\n    <tr>\n      <th>50%</th>\n      <td>4.500000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>3.000000</td>\n      <td>...</td>\n      <td>38.000000</td>\n      <td>32.000000</td>\n      <td>3.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n    </tr>\n    <tr>\n      <th>75%</th>\n      <td>7.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>49.000000</td>\n      <td>91.000000</td>\n      <td>105.000000</td>\n      <td>103.00000</td>\n      <td>102.000000</td>\n      <td>108.000000</td>\n      <td>103.000000</td>\n      <td>77.000000</td>\n      <td>20.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>8.000000</td>\n      <td>100.000000</td>\n      <td>177.000000</td>\n      <td>...</td>\n      <td>201.000000</td>\n      <td>196.000000</td>\n      <td>176.000000</td>\n      <td>135.000000</td>\n      <td>68.000000</td>\n      <td>60.000000</td>\n      <td>70.000000</td>\n      <td>22.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>2.000000</td>\n      <td>43.000000</td>\n      <td>81.000000</td>\n      <td>105.000000</td>\n      <td>108.00000</td>\n      <td>94.000000</td>\n      <td>84.000000</td>\n      <td>91.000000</td>\n      <td>104.000000</td>\n      <td>93.000000</td>\n      <td>58.000000</td>\n      <td>9.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.000000</td>\n      <td>0.00000</td>\n    </tr>\n    <tr>\n      <th>max</th>\n      <td>9.000000</td>\n      <td>16.000000</td>\n      <td>36.000000</td>\n      <td>226.000000</td>\n      <td>164.000000</td>\n      <td>227.000000</td>\n      <td>230.000000</td>\n      <td>224.000000</td>\n      <td>255.000000</td>\n      <td>254.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.00000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>222.000000</td>\n      <td>220.000000</td>\n      <td>239.000000</td>\n      <td>255.000000</td>\n      <td>180.000000</td>\n      <td>189.000000</td>\n      <td>105.000000</td>\n      <td>167.000000</td>\n      <td>219.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>...</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>246.000000</td>\n      <td>81.000000</td>\n      <td>239.000000</td>\n      <td>230.000000</td>\n      <td>252.000000</td>\n      <td>250.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.00000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>255.000000</td>\n      <td>170.00000</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "f7d368cb8071302e9c294e342c9f64c32b306e9b"
      },
      "cell_type": "markdown",
      "source": "from above : Since the image data in x_train and x_test is from 0 to 255 , we need to rescale this from 0 to 1.To do this we need to divide the x_train and x_test by 255\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "41e8809b133177e6d2b1c858f656f53852dc8336"
      },
      "cell_type": "code",
      "source": "x_train = train_data[:,1:]/255\n\ny_train = train_data[:,0]\n\nx_test= test_data[:,1:]/255\n\ny_test=test_data[:,0]\n",
      "execution_count": 24,
      "outputs": []
    },
    {
      "metadata": {
        "_uuid": "d54a00227d0b57742eb8fa7260b41b974ede1707"
      },
      "cell_type": "markdown",
      "source": "Now split the training data into validation and actual training data for training the model and testing it using the validation set. This is achieved using the train_test_split method of scikit learn library.\n\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "d104c22823251e5c293b13676006a41c970fb597"
      },
      "cell_type": "code",
      "source": "x_train,x_validate,y_train,y_validate = train_test_split(x_train,y_train,test_size = 0.2,random_state = 12345)",
      "execution_count": 25,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "491923eb56727200d77963033c21a077bf66efc4"
      },
      "cell_type": "code",
      "source": "#Now let us visualise the sample image how it looks like in 28 * 28 pixel size\n\nimage = x_train[55,:].reshape((28,28))\nplt.imshow(image)\nplt.show()",
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "49156dd359e4c4f94c6a1256a7f01b71f5649f53"
      },
      "cell_type": "markdown",
      "source": "## Creating convolutional Neural Networks\n    1.Define the model\n\n    2.Compile the model\n    \n    3.Fit the model"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "a9d7555318e21064e18e8c20fd60f9671e4a9607"
      },
      "cell_type": "code",
      "source": "# Defined the shape of the image as 3d with rows and columns and 1 for the 3d visualisation\n\nimage_rows = 28\n\nimage_cols = 28\n\nbatch_size = 512\n\nimage_shape = (image_rows,image_cols,1) ",
      "execution_count": 27,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "541eee437b3a698b747a7d67daa74074c5f4298b"
      },
      "cell_type": "code",
      "source": "#formating on the x_train,x_test and x_validate sets.\n\nx_train = x_train.reshape(x_train.shape[0],*image_shape)\nx_test = x_test.reshape(x_test.shape[0],*image_shape)\nx_validate = x_validate.reshape(x_validate.shape[0],*image_shape)",
      "execution_count": 28,
      "outputs": []
    },
    {
      "metadata": {
        "_uuid": "753b4aa6bf227c71618d5c030b5c8c470d36190d"
      },
      "cell_type": "markdown",
      "source": "### Defininig the model"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "1a29d9d92b0fd14fbe2be827aaf9806d5b4b7f7e"
      },
      "cell_type": "code",
      "source": "cnn_model = Sequential([\n    Conv2D(filters=32,kernel_size=3,activation='relu',input_shape = image_shape),\n    MaxPooling2D(pool_size=2) ,# down sampling the output instead of 28*28 it is 14*14\n    Dropout(0.2),\n    Flatten(), # flatten out the layers\n    Dense(32,activation='relu'),\n    Dense(10,activation = 'softmax')\n    \n])",
      "execution_count": 29,
      "outputs": []
    },
    {
      "metadata": {
        "_uuid": "800614e5dcffe6e275263d2cdb6a4b115af57235"
      },
      "cell_type": "markdown",
      "source": "### Compiling model"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "930a82700d5bfc724ea5880c756a3203af059a3c"
      },
      "cell_type": "code",
      "source": "cnn_model.compile(loss ='sparse_categorical_crossentropy', optimizer=Adam(lr=0.001),metrics =['accuracy'])",
      "execution_count": 30,
      "outputs": []
    },
    {
      "metadata": {
        "_uuid": "69bbe422d0ceadc1ff3bc204d4adaa6e69e37931"
      },
      "cell_type": "markdown",
      "source": "### fit the model"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "d66f4f79a62c45698df129b35cea9130fcdf789e"
      },
      "cell_type": "code",
      "source": "history = cnn_model.fit(\n    x_train,\n    y_train,\n    batch_size=batch_size,\n    epochs=50,\n    verbose=1,\n    validation_data=(x_validate,y_validate),\n)",
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Train on 48000 samples, validate on 12000 samples\nEpoch 1/50\n48000/48000 [==============================] - 20s 412us/step - loss: 0.8034 - acc: 0.7339 - val_loss: 0.4955 - val_acc: 0.8250\nEpoch 2/50\n48000/48000 [==============================] - 20s 419us/step - loss: 0.4416 - acc: 0.8429 - val_loss: 0.3973 - val_acc: 0.8642\nEpoch 3/50\n48000/48000 [==============================] - 20s 412us/step - loss: 0.3920 - acc: 0.8623 - val_loss: 0.3647 - val_acc: 0.8728\nEpoch 4/50\n48000/48000 [==============================] - 20s 408us/step - loss: 0.3592 - acc: 0.8744 - val_loss: 0.3461 - val_acc: 0.8813\nEpoch 5/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.3404 - acc: 0.8823 - val_loss: 0.3331 - val_acc: 0.8814\nEpoch 6/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.3234 - acc: 0.8863 - val_loss: 0.3200 - val_acc: 0.8868\nEpoch 7/50\n48000/48000 [==============================] - 19s 404us/step - loss: 0.3080 - acc: 0.8925 - val_loss: 0.3046 - val_acc: 0.8926\nEpoch 8/50\n48000/48000 [==============================] - 19s 402us/step - loss: 0.2969 - acc: 0.8963 - val_loss: 0.3066 - val_acc: 0.8906\nEpoch 9/50\n48000/48000 [==============================] - 19s 404us/step - loss: 0.2898 - acc: 0.8981 - val_loss: 0.2925 - val_acc: 0.8976\nEpoch 10/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.2808 - acc: 0.9013 - val_loss: 0.2876 - val_acc: 0.8967\nEpoch 11/50\n48000/48000 [==============================] - 20s 407us/step - loss: 0.2736 - acc: 0.9046 - val_loss: 0.2766 - val_acc: 0.9011\nEpoch 12/50\n48000/48000 [==============================] - 20s 408us/step - loss: 0.2645 - acc: 0.9066 - val_loss: 0.2813 - val_acc: 0.8981\nEpoch 13/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.2585 - acc: 0.9088 - val_loss: 0.2732 - val_acc: 0.9052\nEpoch 14/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.2549 - acc: 0.9104 - val_loss: 0.2662 - val_acc: 0.9063\nEpoch 15/50\n48000/48000 [==============================] - 19s 405us/step - loss: 0.2456 - acc: 0.9137 - val_loss: 0.2650 - val_acc: 0.9037\nEpoch 16/50\n48000/48000 [==============================] - 20s 410us/step - loss: 0.2433 - acc: 0.9124 - val_loss: 0.2601 - val_acc: 0.9072\nEpoch 17/50\n48000/48000 [==============================] - 20s 409us/step - loss: 0.2367 - acc: 0.9154 - val_loss: 0.2604 - val_acc: 0.9072\nEpoch 18/50\n48000/48000 [==============================] - 20s 423us/step - loss: 0.2338 - acc: 0.9167 - val_loss: 0.2576 - val_acc: 0.9083\nEpoch 19/50\n48000/48000 [==============================] - 20s 409us/step - loss: 0.2301 - acc: 0.9182 - val_loss: 0.2552 - val_acc: 0.9096\nEpoch 20/50\n48000/48000 [==============================] - 19s 405us/step - loss: 0.2263 - acc: 0.9186 - val_loss: 0.2528 - val_acc: 0.9090\nEpoch 21/50\n48000/48000 [==============================] - 19s 403us/step - loss: 0.2190 - acc: 0.9221 - val_loss: 0.2500 - val_acc: 0.9099\nEpoch 22/50\n48000/48000 [==============================] - 19s 404us/step - loss: 0.2154 - acc: 0.9232 - val_loss: 0.2487 - val_acc: 0.9118\nEpoch 23/50\n48000/48000 [==============================] - 19s 402us/step - loss: 0.2133 - acc: 0.9233 - val_loss: 0.2515 - val_acc: 0.9092\nEpoch 24/50\n48000/48000 [==============================] - 19s 403us/step - loss: 0.2078 - acc: 0.9257 - val_loss: 0.2461 - val_acc: 0.9105\nEpoch 25/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.2059 - acc: 0.9271 - val_loss: 0.2601 - val_acc: 0.9087\nEpoch 26/50\n48000/48000 [==============================] - 19s 404us/step - loss: 0.2045 - acc: 0.9267 - val_loss: 0.2438 - val_acc: 0.9115\nEpoch 27/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.1994 - acc: 0.9280 - val_loss: 0.2465 - val_acc: 0.9126\nEpoch 28/50\n48000/48000 [==============================] - 19s 403us/step - loss: 0.1937 - acc: 0.9302 - val_loss: 0.2409 - val_acc: 0.9121\nEpoch 29/50\n48000/48000 [==============================] - 19s 402us/step - loss: 0.1909 - acc: 0.9311 - val_loss: 0.2391 - val_acc: 0.9140\nEpoch 30/50\n48000/48000 [==============================] - 19s 403us/step - loss: 0.1902 - acc: 0.9318 - val_loss: 0.2429 - val_acc: 0.9125\nEpoch 31/50\n48000/48000 [==============================] - 19s 405us/step - loss: 0.1859 - acc: 0.9337 - val_loss: 0.2450 - val_acc: 0.9116\nEpoch 32/50\n48000/48000 [==============================] - 20s 407us/step - loss: 0.1817 - acc: 0.9346 - val_loss: 0.2390 - val_acc: 0.9146\nEpoch 33/50\n48000/48000 [==============================] - 19s 406us/step - loss: 0.1785 - acc: 0.9359 - val_loss: 0.2371 - val_acc: 0.9152\nEpoch 34/50\n48000/48000 [==============================] - 20s 417us/step - loss: 0.1769 - acc: 0.9366 - val_loss: 0.2395 - val_acc: 0.9156\nEpoch 35/50\n48000/48000 [==============================] - 20s 410us/step - loss: 0.1743 - acc: 0.9362 - val_loss: 0.2396 - val_acc: 0.9150\nEpoch 36/50\n48000/48000 [==============================] - 19s 404us/step - loss: 0.1752 - acc: 0.9371 - val_loss: 0.2393 - val_acc: 0.9146\nEpoch 37/50\n48000/48000 [==============================] - 19s 402us/step - loss: 0.1719 - acc: 0.9384 - val_loss: 0.2376 - val_acc: 0.9137\nEpoch 38/50\n48000/48000 [==============================] - 19s 402us/step - loss: 0.1651 - acc: 0.9409 - val_loss: 0.2528 - val_acc: 0.9071\nEpoch 39/50\n48000/48000 [==============================] - 19s 396us/step - loss: 0.1640 - acc: 0.9403 - val_loss: 0.2408 - val_acc: 0.9153\nEpoch 40/50\n48000/48000 [==============================] - 19s 401us/step - loss: 0.1595 - acc: 0.9434 - val_loss: 0.2382 - val_acc: 0.9151\nEpoch 41/50\n48000/48000 [==============================] - 19s 401us/step - loss: 0.1593 - acc: 0.9429 - val_loss: 0.2394 - val_acc: 0.9139\nEpoch 42/50\n48000/48000 [==============================] - 19s 403us/step - loss: 0.1572 - acc: 0.9434 - val_loss: 0.2382 - val_acc: 0.9147\nEpoch 43/50\n48000/48000 [==============================] - 19s 401us/step - loss: 0.1562 - acc: 0.9436 - val_loss: 0.2405 - val_acc: 0.9146\nEpoch 44/50\n48000/48000 [==============================] - 19s 401us/step - loss: 0.1500 - acc: 0.9453 - val_loss: 0.2448 - val_acc: 0.9129\nEpoch 45/50\n48000/48000 [==============================] - 19s 401us/step - loss: 0.1493 - acc: 0.9462 - val_loss: 0.2453 - val_acc: 0.9137\nEpoch 46/50\n48000/48000 [==============================] - 19s 396us/step - loss: 0.1490 - acc: 0.9461 - val_loss: 0.2404 - val_acc: 0.9157\nEpoch 47/50\n48000/48000 [==============================] - 19s 399us/step - loss: 0.1468 - acc: 0.9471 - val_loss: 0.2413 - val_acc: 0.9158\nEpoch 48/50\n48000/48000 [==============================] - 19s 404us/step - loss: 0.1448 - acc: 0.9474 - val_loss: 0.2409 - val_acc: 0.9170\nEpoch 49/50\n48000/48000 [==============================] - 19s 402us/step - loss: 0.1427 - acc: 0.9484 - val_loss: 0.2436 - val_acc: 0.9156\nEpoch 50/50\n48000/48000 [==============================] - 20s 419us/step - loss: 0.1405 - acc: 0.9492 - val_loss: 0.2543 - val_acc: 0.9141\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "81d0ad275058cf64c6a57cf8f076a439ec5d435b"
      },
      "cell_type": "markdown",
      "source": "### Evaluate / score of the model"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "a7d5ee43b3f68a5c11af3a321e0f6ee85e1284d1"
      },
      "cell_type": "code",
      "source": "score = cnn_model.evaluate(x_test,y_test,verbose=0)\nprint('Test Loss : {:.4f}'.format(score[0]))\nprint('Test Accuracy : {:.4f}'.format(score[1]))",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "_uuid": "c3e8b457e05b78b47400f6d96760c78976475e7a"
      },
      "cell_type": "markdown",
      "source": "### plotting the result ie training accuracy vs validation accuracy and training loss and validation loss"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "d2dc15d452cee804ed2027b8f65cdd493ae20144"
      },
      "cell_type": "code",
      "source": "import matplotlib.pyplot as plt\n%matplotlib inline\naccuracy = history.history['acc']\nval_accuracy = history.history['val_acc']\nloss = history.history['loss']\nval_loss = history.history['val_loss']\nepochs = range(len(accuracy))\nplt.plot(epochs, accuracy, 'bo', label='Training Accuracy')\nplt.plot(epochs, val_accuracy, 'b', label='Validation Accuracy')\nplt.title('Training and Validation accuracy')\nplt.legend()\nplt.figure()\nplt.plot(epochs, loss, 'bo', label='Training Loss')\nplt.plot(epochs, val_loss, 'b', label='Validation Loss')\nplt.title('Training and validation loss')\nplt.legend()\nplt.show()",
      "execution_count": 39,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "b041199ed35333ddddc541796f46ee952ec35148"
      },
      "cell_type": "markdown",
      "source": "\n### summarizing the performance of our classifier as follows\n\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "b91f7896d8bd8c5441336ef98ce30750d8877ffc"
      },
      "cell_type": "code",
      "source": "#get the predictions for the test data\npredicted_classes = cnn_model.predict_classes(x_test)\n#get the indices to be plotted\ny_true = test_df.iloc[:, 0]\ncorrect = np.nonzero(predicted_classes==y_true)[0]\nincorrect = np.nonzero(predicted_classes!=y_true)[0]\nfrom sklearn.metrics import classification_report\ntarget_names = [\"Class {}\".format(i) for i in range(num_classes)]\nprint(classification_report(y_true, predicted_classes, target_names=target_names))",
      "execution_count": 40,
      "outputs": [
        {
          "output_type": "stream",
          "text": "             precision    recall  f1-score   support\n\n    Class 0       0.79      0.92      0.85      1000\n    Class 1       0.98      0.98      0.98      1000\n    Class 2       0.85      0.88      0.87      1000\n    Class 3       0.92      0.93      0.93      1000\n    Class 4       0.87      0.88      0.88      1000\n    Class 5       0.98      0.98      0.98      1000\n    Class 6       0.83      0.65      0.73      1000\n    Class 7       0.96      0.96      0.96      1000\n    Class 8       0.98      0.98      0.98      1000\n    Class 9       0.96      0.97      0.97      1000\n\navg / total       0.91      0.91      0.91     10000\n\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "355aadb864ac3dde8f570cb8ffa8a9936cbb7252"
      },
      "cell_type": "markdown",
      "source": "It's apparent that our classifier is underperforming for class 6 in terms of both precision and recall. For class 2, classifier is slightly lacking precision whereas it is slightly lacking recall (i.e. missed) for class 4.\n\n\nPerhaps we would gain more insight after visualizing the correct and incorrect predictions.\n"
    },
    {
      "metadata": {
        "_uuid": "51648cc452989a3e046a94a6194ecd9bd009c31e"
      },
      "cell_type": "markdown",
      "source": "\n### Subset of correctly predicted classes."
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "9ab87b2675b6e83764ab9e3499e7c7d823db351a"
      },
      "cell_type": "code",
      "source": "for i, correct in enumerate(correct[:9]):\n    plt.subplot(3,3,i+1)\n    plt.imshow(x_test[correct].reshape(28,28), cmap='gray', interpolation='none')\n    plt.title(\"Predicted {}, Class {}\".format(predicted_classes[correct], y_true[correct]))\n    plt.tight_layout()\n",
      "execution_count": 41,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 9 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "f395d5c26261146a3b8cfeebf6e21723c7175df4"
      },
      "cell_type": "markdown",
      "source": "### Subset of incorrectly predicted classes.\n"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "8416e396167deeb5ce6948038ec82f92e0efa571"
      },
      "cell_type": "code",
      "source": "\nfor i, incorrect in enumerate(incorrect[0:9]):\n    plt.subplot(3,3,i+1)\n    plt.imshow(x_test[incorrect].reshape(28,28), cmap='gray', interpolation='none')\n    plt.title(\"Predicted {}, Class {}\".format(predicted_classes[incorrect], y_true[incorrect]))\n    plt.tight_layout()",
      "execution_count": 42,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 9 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "metadata": {
        "_uuid": "fc7620b512ef52baa094407a1b8b8ac1b85b007e"
      },
      "cell_type": "markdown",
      "source": "### Predict and Submit results "
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "e65a21ff2afae6b618f6ff43dc86fa2e882b4281"
      },
      "cell_type": "code",
      "source": "\n\n# predict results\n \nresults = cnn_model.predict(x_test)\n\n# select the indix with the maximum probability\n \nresults = np.argmax(results,axis = 1)\n\nresults = pd.Series(results,name=\"Label\")\n\n\n\nsubmission = pd.concat([pd.Series(range(1,10001),name = \"id\"),results],axis = 1)\n\nsubmission.to_csv(\"cnn_mnist_datagen.csv\",index=False)\n",
      "execution_count": 44,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "02ad36fdd318f9d91988bd4e7bcc1f90dbf68143"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "125b8a6086d38b077990478b76c5ca04c7fba4a8"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "18667b348d8c72d611cf9a94744ce6761dda834f"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.6.6",
      "mimetype": "text/x-python",
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "pygments_lexer": "ipython3",
      "nbconvert_exporter": "python",
      "file_extension": ".py"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 1
}