MATH465_Project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#  Importing libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['datasets_54339_104884_HAM10000_metadata.csv', 'hmnist_28_28_L.csv', 'hmnist_28_28_RGB.csv']\n"
     ]
    }
   ],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "from keras.models import Sequential,Model,load_model\n",
    "from keras.layers import Dense, Dropout, Flatten, Conv2D, MaxPool2D,GlobalAveragePooling2D\n",
    "from keras.optimizers import SGD, Adam\n",
    "from keras.utils import to_categorical\n",
    "from keras import applications\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "import sklearn.metrics as metrics\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "import cv2\n",
    "import glob\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "import os\n",
    "print(os.listdir(\"C:\\\\Users\\\\GERMAN\\\\Documents\\\\bilkent 4\\\\spring\\\\math 465 data science\\\\project\\\\dataset\\\\data\"))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Importing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_csv('C:\\\\Users\\\\GERMAN\\\\Documents\\\\bilkent 4\\\\spring\\\\math 465 data science\\\\project\\\\dataset\\\\data\\\\hmnist_28_28_RGB.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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>pixel0000</th>\n",
       "      <th>pixel0001</th>\n",
       "      <th>pixel0002</th>\n",
       "      <th>pixel0003</th>\n",
       "      <th>pixel0004</th>\n",
       "      <th>pixel0005</th>\n",
       "      <th>pixel0006</th>\n",
       "      <th>pixel0007</th>\n",
       "      <th>pixel0008</th>\n",
       "      <th>pixel0009</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel2343</th>\n",
       "      <th>pixel2344</th>\n",
       "      <th>pixel2345</th>\n",
       "      <th>pixel2346</th>\n",
       "      <th>pixel2347</th>\n",
       "      <th>pixel2348</th>\n",
       "      <th>pixel2349</th>\n",
       "      <th>pixel2350</th>\n",
       "      <th>pixel2351</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>192</td>\n",
       "      <td>153</td>\n",
       "      <td>193</td>\n",
       "      <td>195</td>\n",
       "      <td>155</td>\n",
       "      <td>192</td>\n",
       "      <td>197</td>\n",
       "      <td>154</td>\n",
       "      <td>185</td>\n",
       "      <td>202</td>\n",
       "      <td>...</td>\n",
       "      <td>173</td>\n",
       "      <td>124</td>\n",
       "      <td>138</td>\n",
       "      <td>183</td>\n",
       "      <td>147</td>\n",
       "      <td>166</td>\n",
       "      <td>185</td>\n",
       "      <td>154</td>\n",
       "      <td>177</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>25</td>\n",
       "      <td>14</td>\n",
       "      <td>30</td>\n",
       "      <td>68</td>\n",
       "      <td>48</td>\n",
       "      <td>75</td>\n",
       "      <td>123</td>\n",
       "      <td>93</td>\n",
       "      <td>126</td>\n",
       "      <td>158</td>\n",
       "      <td>...</td>\n",
       "      <td>60</td>\n",
       "      <td>39</td>\n",
       "      <td>55</td>\n",
       "      <td>25</td>\n",
       "      <td>14</td>\n",
       "      <td>28</td>\n",
       "      <td>25</td>\n",
       "      <td>14</td>\n",
       "      <td>27</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>192</td>\n",
       "      <td>138</td>\n",
       "      <td>153</td>\n",
       "      <td>200</td>\n",
       "      <td>145</td>\n",
       "      <td>163</td>\n",
       "      <td>201</td>\n",
       "      <td>142</td>\n",
       "      <td>160</td>\n",
       "      <td>206</td>\n",
       "      <td>...</td>\n",
       "      <td>167</td>\n",
       "      <td>129</td>\n",
       "      <td>143</td>\n",
       "      <td>159</td>\n",
       "      <td>124</td>\n",
       "      <td>142</td>\n",
       "      <td>136</td>\n",
       "      <td>104</td>\n",
       "      <td>117</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>38</td>\n",
       "      <td>19</td>\n",
       "      <td>30</td>\n",
       "      <td>95</td>\n",
       "      <td>59</td>\n",
       "      <td>72</td>\n",
       "      <td>143</td>\n",
       "      <td>103</td>\n",
       "      <td>119</td>\n",
       "      <td>171</td>\n",
       "      <td>...</td>\n",
       "      <td>44</td>\n",
       "      <td>26</td>\n",
       "      <td>36</td>\n",
       "      <td>25</td>\n",
       "      <td>12</td>\n",
       "      <td>17</td>\n",
       "      <td>25</td>\n",
       "      <td>12</td>\n",
       "      <td>15</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>158</td>\n",
       "      <td>113</td>\n",
       "      <td>139</td>\n",
       "      <td>194</td>\n",
       "      <td>144</td>\n",
       "      <td>174</td>\n",
       "      <td>215</td>\n",
       "      <td>162</td>\n",
       "      <td>191</td>\n",
       "      <td>225</td>\n",
       "      <td>...</td>\n",
       "      <td>209</td>\n",
       "      <td>166</td>\n",
       "      <td>185</td>\n",
       "      <td>172</td>\n",
       "      <td>135</td>\n",
       "      <td>149</td>\n",
       "      <td>109</td>\n",
       "      <td>78</td>\n",
       "      <td>92</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 2353 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   pixel0000  pixel0001  pixel0002  pixel0003  pixel0004  pixel0005  \\\n",
       "0        192        153        193        195        155        192   \n",
       "1         25         14         30         68         48         75   \n",
       "2        192        138        153        200        145        163   \n",
       "3         38         19         30         95         59         72   \n",
       "4        158        113        139        194        144        174   \n",
       "\n",
       "   pixel0006  pixel0007  pixel0008  pixel0009  ...    pixel2343  pixel2344  \\\n",
       "0        197        154        185        202  ...          173        124   \n",
       "1        123         93        126        158  ...           60         39   \n",
       "2        201        142        160        206  ...          167        129   \n",
       "3        143        103        119        171  ...           44         26   \n",
       "4        215        162        191        225  ...          209        166   \n",
       "\n",
       "   pixel2345  pixel2346  pixel2347  pixel2348  pixel2349  pixel2350  \\\n",
       "0        138        183        147        166        185        154   \n",
       "1         55         25         14         28         25         14   \n",
       "2        143        159        124        142        136        104   \n",
       "3         36         25         12         17         25         12   \n",
       "4        185        172        135        149        109         78   \n",
       "\n",
       "   pixel2351  label  \n",
       "0        177      2  \n",
       "1         27      2  \n",
       "2        117      2  \n",
       "3         15      2  \n",
       "4         92      2  \n",
       "\n",
       "[5 rows x 2353 columns]"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1de8b97f940>"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax1 = plt.subplots(1, 1, figsize = (10, 5))\n",
    "sns.countplot(y='label',data=data, hue=\"label\",ax=ax1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Importing metadata"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cases include a representative collection of all important diagnostic categories in the realm of pigmented lesions: Actinic keratoses and intraepithelial carcinoma / Bowen's disease (akiec), basal cell carcinoma (bcc), benign keratosis-like lesions (solar lentigines / seborrheic keratoses and lichen-planus like keratoses, bkl), dermatofibroma (df), melanoma (mel), melanocytic nevi (nv) and vascular lesions (angiomas, angiokeratomas, pyogenic granulomas and hemorrhage, vasc).\n",
    "\n",
    "More than 50% of lesions are confirmed through histopathology (histo), the ground truth for the rest of the cases is either follow-up examination (followup), expert consensus (consensus), or confirmation by in-vivo confocal microscopy (confocal). "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "metadata = pd.read_csv('C:\\\\Users\\\\GERMAN\\\\Documents\\\\bilkent 4\\\\spring\\\\math 465 data science\\\\project\\\\dataset\\\\data\\\\datasets_54339_104884_HAM10000_metadata.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "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>lesion_id</th>\n",
       "      <th>image_id</th>\n",
       "      <th>dx</th>\n",
       "      <th>dx_type</th>\n",
       "      <th>age</th>\n",
       "      <th>sex</th>\n",
       "      <th>localization</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>HAM_0000118</td>\n",
       "      <td>ISIC_0027419</td>\n",
       "      <td>bkl</td>\n",
       "      <td>histo</td>\n",
       "      <td>80.0</td>\n",
       "      <td>male</td>\n",
       "      <td>scalp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>HAM_0000118</td>\n",
       "      <td>ISIC_0025030</td>\n",
       "      <td>bkl</td>\n",
       "      <td>histo</td>\n",
       "      <td>80.0</td>\n",
       "      <td>male</td>\n",
       "      <td>scalp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>HAM_0002730</td>\n",
       "      <td>ISIC_0026769</td>\n",
       "      <td>bkl</td>\n",
       "      <td>histo</td>\n",
       "      <td>80.0</td>\n",
       "      <td>male</td>\n",
       "      <td>scalp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>HAM_0002730</td>\n",
       "      <td>ISIC_0025661</td>\n",
       "      <td>bkl</td>\n",
       "      <td>histo</td>\n",
       "      <td>80.0</td>\n",
       "      <td>male</td>\n",
       "      <td>scalp</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>HAM_0001466</td>\n",
       "      <td>ISIC_0031633</td>\n",
       "      <td>bkl</td>\n",
       "      <td>histo</td>\n",
       "      <td>75.0</td>\n",
       "      <td>male</td>\n",
       "      <td>ear</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     lesion_id      image_id   dx dx_type   age   sex localization\n",
       "0  HAM_0000118  ISIC_0027419  bkl   histo  80.0  male        scalp\n",
       "1  HAM_0000118  ISIC_0025030  bkl   histo  80.0  male        scalp\n",
       "2  HAM_0002730  ISIC_0026769  bkl   histo  80.0  male        scalp\n",
       "3  HAM_0002730  ISIC_0025661  bkl   histo  80.0  male        scalp\n",
       "4  HAM_0001466  ISIC_0031633  bkl   histo  75.0  male          ear"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "metadata.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Dropping the label column from the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "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>pixel0000</th>\n",
       "      <th>pixel0001</th>\n",
       "      <th>pixel0002</th>\n",
       "      <th>pixel0003</th>\n",
       "      <th>pixel0004</th>\n",
       "      <th>pixel0005</th>\n",
       "      <th>pixel0006</th>\n",
       "      <th>pixel0007</th>\n",
       "      <th>pixel0008</th>\n",
       "      <th>pixel0009</th>\n",
       "      <th>...</th>\n",
       "      <th>pixel2342</th>\n",
       "      <th>pixel2343</th>\n",
       "      <th>pixel2344</th>\n",
       "      <th>pixel2345</th>\n",
       "      <th>pixel2346</th>\n",
       "      <th>pixel2347</th>\n",
       "      <th>pixel2348</th>\n",
       "      <th>pixel2349</th>\n",
       "      <th>pixel2350</th>\n",
       "      <th>pixel2351</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>192</td>\n",
       "      <td>153</td>\n",
       "      <td>193</td>\n",
       "      <td>195</td>\n",
       "      <td>155</td>\n",
       "      <td>192</td>\n",
       "      <td>197</td>\n",
       "      <td>154</td>\n",
       "      <td>185</td>\n",
       "      <td>202</td>\n",
       "      <td>...</td>\n",
       "      <td>134</td>\n",
       "      <td>173</td>\n",
       "      <td>124</td>\n",
       "      <td>138</td>\n",
       "      <td>183</td>\n",
       "      <td>147</td>\n",
       "      <td>166</td>\n",
       "      <td>185</td>\n",
       "      <td>154</td>\n",
       "      <td>177</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>25</td>\n",
       "      <td>14</td>\n",
       "      <td>30</td>\n",
       "      <td>68</td>\n",
       "      <td>48</td>\n",
       "      <td>75</td>\n",
       "      <td>123</td>\n",
       "      <td>93</td>\n",
       "      <td>126</td>\n",
       "      <td>158</td>\n",
       "      <td>...</td>\n",
       "      <td>82</td>\n",
       "      <td>60</td>\n",
       "      <td>39</td>\n",
       "      <td>55</td>\n",
       "      <td>25</td>\n",
       "      <td>14</td>\n",
       "      <td>28</td>\n",
       "      <td>25</td>\n",
       "      <td>14</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>192</td>\n",
       "      <td>138</td>\n",
       "      <td>153</td>\n",
       "      <td>200</td>\n",
       "      <td>145</td>\n",
       "      <td>163</td>\n",
       "      <td>201</td>\n",
       "      <td>142</td>\n",
       "      <td>160</td>\n",
       "      <td>206</td>\n",
       "      <td>...</td>\n",
       "      <td>149</td>\n",
       "      <td>167</td>\n",
       "      <td>129</td>\n",
       "      <td>143</td>\n",
       "      <td>159</td>\n",
       "      <td>124</td>\n",
       "      <td>142</td>\n",
       "      <td>136</td>\n",
       "      <td>104</td>\n",
       "      <td>117</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>38</td>\n",
       "      <td>19</td>\n",
       "      <td>30</td>\n",
       "      <td>95</td>\n",
       "      <td>59</td>\n",
       "      <td>72</td>\n",
       "      <td>143</td>\n",
       "      <td>103</td>\n",
       "      <td>119</td>\n",
       "      <td>171</td>\n",
       "      <td>...</td>\n",
       "      <td>73</td>\n",
       "      <td>44</td>\n",
       "      <td>26</td>\n",
       "      <td>36</td>\n",
       "      <td>25</td>\n",
       "      <td>12</td>\n",
       "      <td>17</td>\n",
       "      <td>25</td>\n",
       "      <td>12</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>158</td>\n",
       "      <td>113</td>\n",
       "      <td>139</td>\n",
       "      <td>194</td>\n",
       "      <td>144</td>\n",
       "      <td>174</td>\n",
       "      <td>215</td>\n",
       "      <td>162</td>\n",
       "      <td>191</td>\n",
       "      <td>225</td>\n",
       "      <td>...</td>\n",
       "      <td>201</td>\n",
       "      <td>209</td>\n",
       "      <td>166</td>\n",
       "      <td>185</td>\n",
       "      <td>172</td>\n",
       "      <td>135</td>\n",
       "      <td>149</td>\n",
       "      <td>109</td>\n",
       "      <td>78</td>\n",
       "      <td>92</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 2352 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   pixel0000  pixel0001  pixel0002  pixel0003  pixel0004  pixel0005  \\\n",
       "0        192        153        193        195        155        192   \n",
       "1         25         14         30         68         48         75   \n",
       "2        192        138        153        200        145        163   \n",
       "3         38         19         30         95         59         72   \n",
       "4        158        113        139        194        144        174   \n",
       "\n",
       "   pixel0006  pixel0007  pixel0008  pixel0009    ...      pixel2342  \\\n",
       "0        197        154        185        202    ...            134   \n",
       "1        123         93        126        158    ...             82   \n",
       "2        201        142        160        206    ...            149   \n",
       "3        143        103        119        171    ...             73   \n",
       "4        215        162        191        225    ...            201   \n",
       "\n",
       "   pixel2343  pixel2344  pixel2345  pixel2346  pixel2347  pixel2348  \\\n",
       "0        173        124        138        183        147        166   \n",
       "1         60         39         55         25         14         28   \n",
       "2        167        129        143        159        124        142   \n",
       "3         44         26         36         25         12         17   \n",
       "4        209        166        185        172        135        149   \n",
       "\n",
       "   pixel2349  pixel2350  pixel2351  \n",
       "0        185        154        177  \n",
       "1         25         14         27  \n",
       "2        136        104        117  \n",
       "3         25         12         15  \n",
       "4        109         78         92  \n",
       "\n",
       "[5 rows x 2352 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Y = data['label']\n",
    "data.drop('label',axis=1,inplace=True)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Label encoding. We have 7 classes for classification."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y = to_categorical(Y,num_classes=7)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Splitting data to training data and testing data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "# Test_size should be between 0.0 and 1.0 and represent the proportion of the dataset to include in the train split\n",
    "# if you use random_state=some_number, then you can guarantee that the output of Run 1 will be equal to the output of Run 2,\n",
    "# i.e. your split will be always the same.\n",
    "X_train,X_test,Y_train,Y_test = train_test_split(data,Y,test_size=0.2,random_state=42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### TODO: Oversampling minority classes to have a more balanced data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number of training examples = 8012\n",
      "number of test examples = 2003\n",
      "X_train shape: (8012, 2352)\n",
      "Y_train shape: (8012, 7)\n",
      "X_test shape: (2003, 2352)\n",
      "Y_test shape: (2003, 7)\n"
     ]
    }
   ],
   "source": [
    "print (\"number of training examples = \" + str(X_train.shape[0]))\n",
    "print (\"number of test examples = \" + str(X_test.shape[0]))\n",
    "print (\"X_train shape: \" + str(X_train.shape))\n",
    "print (\"Y_train shape: \" + str(Y_train.shape))\n",
    "print (\"X_test shape: \" + str(X_test.shape))\n",
    "print (\"Y_test shape: \" + str(Y_test.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Reshaping to X_train and X_test to 3D arrays"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X_train shape: (8012, 28, 28, 3)\n",
      "X_test shape: (2003, 28, 28, 3)\n"
     ]
    }
   ],
   "source": [
    "X_train = X_train.values\n",
    "X_train = np.reshape(X_train, (X_train.shape[0], 28, 28, 3))\n",
    "\n",
    "X_test = X_test.values\n",
    "X_test = np.reshape(X_test, (X_test.shape[0], 28, 28, 3))\n",
    "\n",
    "print (\"X_train shape: \" + str(X_train.shape))\n",
    "print (\"X_test shape: \" + str(X_test.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Padding the X_train and X_test image arrays with zeros so that it becomes an image array of size 32x32x3, because this is the size of the minimum input of ResNet model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "hpad = np.zeros(( 28, 2 , 3))\n",
    "vpad = np.zeros(( 2, 32, 3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "#im = np.asarray([np.pad(x, (2,), 'constant', constant_values=(0 ,)) for x in X_train])\n",
    "X_train = np.asarray([np.hstack((hpad, x, hpad)) for x in X_train])\n",
    "X_train = np.asarray([np.vstack((vpad, x, vpad)) for x in X_train])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(8012, 32, 32, 3)\n"
     ]
    }
   ],
   "source": [
    "print(str(X_train.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2003, 32, 32, 3)\n"
     ]
    }
   ],
   "source": [
    "X_test = np.asarray([np.hstack((hpad, x, hpad)) for x in X_test])\n",
    "X_test = np.asarray([np.vstack((vpad, x, vpad)) for x in X_test])\n",
    "print(str(X_test.shape))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# RESNET-50"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Import the ResNet model for leraning. Here weights=None since we want to initialize the model with random weights, otherwise we can also load the pretrained ImageNet weights. I set include_top=False to not include the final pooling and fully connected layer in the original model. We add Global Average Pooling and a dense output layer to the ResNet-50 model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "img_height,img_width = 32,32 \n",
    "num_classes = 7"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\GERMAN\\Anaconda3\\lib\\site-packages\\keras_applications\\resnet50.py:265: UserWarning: The output shape of `ResNet50(include_top=False)` has been changed since Keras 2.2.0.\n",
      "  warnings.warn('The output shape of `ResNet50(include_top=False)` '\n"
     ]
    }
   ],
   "source": [
    "#If imagenet weights are being loaded, \n",
    "#input must have a static square shape (one of (128, 128), (160, 160), (192, 192), or (224, 224))\n",
    "base_model = applications.resnet50.ResNet50(weights= None, include_top=False, input_shape= (img_height,img_width,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = base_model.output\n",
    "x = GlobalAveragePooling2D()(x)\n",
    "x = Dropout(0.7)(x)\n",
    "predictions = Dense(num_classes, activation= 'softmax')(x)\n",
    "model = Model(inputs = base_model.input, outputs = predictions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We use Adam optimizer,learning rate= 0.001 and 'categorical crossentropy' loss function. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sgd = SGD(lr=lrate, momentum=0.9, decay=decay, nesterov=False)\n",
    "adam = Adam(lr=0.001)\n",
    "model.compile(optimizer= adam, loss='categorical_crossentropy', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_3 (InputLayer)            (None, 32, 32, 3)    0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1_pad (ZeroPadding2D)       (None, 38, 38, 3)    0           input_3[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1 (Conv2D)                  (None, 16, 16, 64)   9472        conv1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "bn_conv1 (BatchNormalization)   (None, 16, 16, 64)   256         conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "activation_99 (Activation)      (None, 16, 16, 64)   0           bn_conv1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "pool1_pad (ZeroPadding2D)       (None, 18, 18, 64)   0           activation_99[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2D)  (None, 8, 8, 64)     0           pool1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2a (Conv2D)         (None, 8, 8, 64)     4160        max_pooling2d_3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2a (BatchNormalizati (None, 8, 8, 64)     256         res2a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_100 (Activation)     (None, 8, 8, 64)     0           bn2a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2b (Conv2D)         (None, 8, 8, 64)     36928       activation_100[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2b (BatchNormalizati (None, 8, 8, 64)     256         res2a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_101 (Activation)     (None, 8, 8, 64)     0           bn2a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2c (Conv2D)         (None, 8, 8, 256)    16640       activation_101[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch1 (Conv2D)          (None, 8, 8, 256)    16640       max_pooling2d_3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2c (BatchNormalizati (None, 8, 8, 256)    1024        res2a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch1 (BatchNormalizatio (None, 8, 8, 256)    1024        res2a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_33 (Add)                    (None, 8, 8, 256)    0           bn2a_branch2c[0][0]              \n",
      "                                                                 bn2a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_102 (Activation)     (None, 8, 8, 256)    0           add_33[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2a (Conv2D)         (None, 8, 8, 64)     16448       activation_102[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2a (BatchNormalizati (None, 8, 8, 64)     256         res2b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_103 (Activation)     (None, 8, 8, 64)     0           bn2b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2b (Conv2D)         (None, 8, 8, 64)     36928       activation_103[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2b (BatchNormalizati (None, 8, 8, 64)     256         res2b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_104 (Activation)     (None, 8, 8, 64)     0           bn2b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2c (Conv2D)         (None, 8, 8, 256)    16640       activation_104[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2c (BatchNormalizati (None, 8, 8, 256)    1024        res2b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_34 (Add)                    (None, 8, 8, 256)    0           bn2b_branch2c[0][0]              \n",
      "                                                                 activation_102[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_105 (Activation)     (None, 8, 8, 256)    0           add_34[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2a (Conv2D)         (None, 8, 8, 64)     16448       activation_105[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2a (BatchNormalizati (None, 8, 8, 64)     256         res2c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_106 (Activation)     (None, 8, 8, 64)     0           bn2c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2b (Conv2D)         (None, 8, 8, 64)     36928       activation_106[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2b (BatchNormalizati (None, 8, 8, 64)     256         res2c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_107 (Activation)     (None, 8, 8, 64)     0           bn2c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2c (Conv2D)         (None, 8, 8, 256)    16640       activation_107[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2c (BatchNormalizati (None, 8, 8, 256)    1024        res2c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_35 (Add)                    (None, 8, 8, 256)    0           bn2c_branch2c[0][0]              \n",
      "                                                                 activation_105[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_108 (Activation)     (None, 8, 8, 256)    0           add_35[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2a (Conv2D)         (None, 4, 4, 128)    32896       activation_108[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2a (BatchNormalizati (None, 4, 4, 128)    512         res3a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_109 (Activation)     (None, 4, 4, 128)    0           bn3a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2b (Conv2D)         (None, 4, 4, 128)    147584      activation_109[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2b (BatchNormalizati (None, 4, 4, 128)    512         res3a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_110 (Activation)     (None, 4, 4, 128)    0           bn3a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2c (Conv2D)         (None, 4, 4, 512)    66048       activation_110[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch1 (Conv2D)          (None, 4, 4, 512)    131584      activation_108[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2c (BatchNormalizati (None, 4, 4, 512)    2048        res3a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch1 (BatchNormalizatio (None, 4, 4, 512)    2048        res3a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_36 (Add)                    (None, 4, 4, 512)    0           bn3a_branch2c[0][0]              \n",
      "                                                                 bn3a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_111 (Activation)     (None, 4, 4, 512)    0           add_36[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2a (Conv2D)         (None, 4, 4, 128)    65664       activation_111[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2a (BatchNormalizati (None, 4, 4, 128)    512         res3b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_112 (Activation)     (None, 4, 4, 128)    0           bn3b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2b (Conv2D)         (None, 4, 4, 128)    147584      activation_112[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2b (BatchNormalizati (None, 4, 4, 128)    512         res3b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_113 (Activation)     (None, 4, 4, 128)    0           bn3b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2c (Conv2D)         (None, 4, 4, 512)    66048       activation_113[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2c (BatchNormalizati (None, 4, 4, 512)    2048        res3b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_37 (Add)                    (None, 4, 4, 512)    0           bn3b_branch2c[0][0]              \n",
      "                                                                 activation_111[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_114 (Activation)     (None, 4, 4, 512)    0           add_37[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2a (Conv2D)         (None, 4, 4, 128)    65664       activation_114[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2a (BatchNormalizati (None, 4, 4, 128)    512         res3c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_115 (Activation)     (None, 4, 4, 128)    0           bn3c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2b (Conv2D)         (None, 4, 4, 128)    147584      activation_115[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2b (BatchNormalizati (None, 4, 4, 128)    512         res3c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_116 (Activation)     (None, 4, 4, 128)    0           bn3c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2c (Conv2D)         (None, 4, 4, 512)    66048       activation_116[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2c (BatchNormalizati (None, 4, 4, 512)    2048        res3c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_38 (Add)                    (None, 4, 4, 512)    0           bn3c_branch2c[0][0]              \n",
      "                                                                 activation_114[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_117 (Activation)     (None, 4, 4, 512)    0           add_38[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2a (Conv2D)         (None, 4, 4, 128)    65664       activation_117[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2a (BatchNormalizati (None, 4, 4, 128)    512         res3d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_118 (Activation)     (None, 4, 4, 128)    0           bn3d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2b (Conv2D)         (None, 4, 4, 128)    147584      activation_118[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2b (BatchNormalizati (None, 4, 4, 128)    512         res3d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_119 (Activation)     (None, 4, 4, 128)    0           bn3d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2c (Conv2D)         (None, 4, 4, 512)    66048       activation_119[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2c (BatchNormalizati (None, 4, 4, 512)    2048        res3d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_39 (Add)                    (None, 4, 4, 512)    0           bn3d_branch2c[0][0]              \n",
      "                                                                 activation_117[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_120 (Activation)     (None, 4, 4, 512)    0           add_39[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2a (Conv2D)         (None, 2, 2, 256)    131328      activation_120[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2a (BatchNormalizati (None, 2, 2, 256)    1024        res4a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_121 (Activation)     (None, 2, 2, 256)    0           bn4a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2b (Conv2D)         (None, 2, 2, 256)    590080      activation_121[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2b (BatchNormalizati (None, 2, 2, 256)    1024        res4a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_122 (Activation)     (None, 2, 2, 256)    0           bn4a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2c (Conv2D)         (None, 2, 2, 1024)   263168      activation_122[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch1 (Conv2D)          (None, 2, 2, 1024)   525312      activation_120[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2c (BatchNormalizati (None, 2, 2, 1024)   4096        res4a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch1 (BatchNormalizatio (None, 2, 2, 1024)   4096        res4a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_40 (Add)                    (None, 2, 2, 1024)   0           bn4a_branch2c[0][0]              \n",
      "                                                                 bn4a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_123 (Activation)     (None, 2, 2, 1024)   0           add_40[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2a (Conv2D)         (None, 2, 2, 256)    262400      activation_123[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2a (BatchNormalizati (None, 2, 2, 256)    1024        res4b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_124 (Activation)     (None, 2, 2, 256)    0           bn4b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2b (Conv2D)         (None, 2, 2, 256)    590080      activation_124[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2b (BatchNormalizati (None, 2, 2, 256)    1024        res4b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_125 (Activation)     (None, 2, 2, 256)    0           bn4b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2c (Conv2D)         (None, 2, 2, 1024)   263168      activation_125[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2c (BatchNormalizati (None, 2, 2, 1024)   4096        res4b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_41 (Add)                    (None, 2, 2, 1024)   0           bn4b_branch2c[0][0]              \n",
      "                                                                 activation_123[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_126 (Activation)     (None, 2, 2, 1024)   0           add_41[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2a (Conv2D)         (None, 2, 2, 256)    262400      activation_126[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2a (BatchNormalizati (None, 2, 2, 256)    1024        res4c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_127 (Activation)     (None, 2, 2, 256)    0           bn4c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2b (Conv2D)         (None, 2, 2, 256)    590080      activation_127[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2b (BatchNormalizati (None, 2, 2, 256)    1024        res4c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_128 (Activation)     (None, 2, 2, 256)    0           bn4c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2c (Conv2D)         (None, 2, 2, 1024)   263168      activation_128[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2c (BatchNormalizati (None, 2, 2, 1024)   4096        res4c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_42 (Add)                    (None, 2, 2, 1024)   0           bn4c_branch2c[0][0]              \n",
      "                                                                 activation_126[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_129 (Activation)     (None, 2, 2, 1024)   0           add_42[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2a (Conv2D)         (None, 2, 2, 256)    262400      activation_129[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2a (BatchNormalizati (None, 2, 2, 256)    1024        res4d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_130 (Activation)     (None, 2, 2, 256)    0           bn4d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2b (Conv2D)         (None, 2, 2, 256)    590080      activation_130[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2b (BatchNormalizati (None, 2, 2, 256)    1024        res4d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_131 (Activation)     (None, 2, 2, 256)    0           bn4d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2c (Conv2D)         (None, 2, 2, 1024)   263168      activation_131[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2c (BatchNormalizati (None, 2, 2, 1024)   4096        res4d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_43 (Add)                    (None, 2, 2, 1024)   0           bn4d_branch2c[0][0]              \n",
      "                                                                 activation_129[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_132 (Activation)     (None, 2, 2, 1024)   0           add_43[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2a (Conv2D)         (None, 2, 2, 256)    262400      activation_132[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2a (BatchNormalizati (None, 2, 2, 256)    1024        res4e_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_133 (Activation)     (None, 2, 2, 256)    0           bn4e_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2b (Conv2D)         (None, 2, 2, 256)    590080      activation_133[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2b (BatchNormalizati (None, 2, 2, 256)    1024        res4e_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_134 (Activation)     (None, 2, 2, 256)    0           bn4e_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2c (Conv2D)         (None, 2, 2, 1024)   263168      activation_134[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2c (BatchNormalizati (None, 2, 2, 1024)   4096        res4e_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_44 (Add)                    (None, 2, 2, 1024)   0           bn4e_branch2c[0][0]              \n",
      "                                                                 activation_132[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_135 (Activation)     (None, 2, 2, 1024)   0           add_44[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2a (Conv2D)         (None, 2, 2, 256)    262400      activation_135[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2a (BatchNormalizati (None, 2, 2, 256)    1024        res4f_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_136 (Activation)     (None, 2, 2, 256)    0           bn4f_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2b (Conv2D)         (None, 2, 2, 256)    590080      activation_136[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2b (BatchNormalizati (None, 2, 2, 256)    1024        res4f_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_137 (Activation)     (None, 2, 2, 256)    0           bn4f_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2c (Conv2D)         (None, 2, 2, 1024)   263168      activation_137[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2c (BatchNormalizati (None, 2, 2, 1024)   4096        res4f_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_45 (Add)                    (None, 2, 2, 1024)   0           bn4f_branch2c[0][0]              \n",
      "                                                                 activation_135[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_138 (Activation)     (None, 2, 2, 1024)   0           add_45[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2a (Conv2D)         (None, 1, 1, 512)    524800      activation_138[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2a (BatchNormalizati (None, 1, 1, 512)    2048        res5a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_139 (Activation)     (None, 1, 1, 512)    0           bn5a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2b (Conv2D)         (None, 1, 1, 512)    2359808     activation_139[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2b (BatchNormalizati (None, 1, 1, 512)    2048        res5a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_140 (Activation)     (None, 1, 1, 512)    0           bn5a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2c (Conv2D)         (None, 1, 1, 2048)   1050624     activation_140[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch1 (Conv2D)          (None, 1, 1, 2048)   2099200     activation_138[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2c (BatchNormalizati (None, 1, 1, 2048)   8192        res5a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch1 (BatchNormalizatio (None, 1, 1, 2048)   8192        res5a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_46 (Add)                    (None, 1, 1, 2048)   0           bn5a_branch2c[0][0]              \n",
      "                                                                 bn5a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_141 (Activation)     (None, 1, 1, 2048)   0           add_46[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2a (Conv2D)         (None, 1, 1, 512)    1049088     activation_141[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2a (BatchNormalizati (None, 1, 1, 512)    2048        res5b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_142 (Activation)     (None, 1, 1, 512)    0           bn5b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2b (Conv2D)         (None, 1, 1, 512)    2359808     activation_142[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2b (BatchNormalizati (None, 1, 1, 512)    2048        res5b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_143 (Activation)     (None, 1, 1, 512)    0           bn5b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2c (Conv2D)         (None, 1, 1, 2048)   1050624     activation_143[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2c (BatchNormalizati (None, 1, 1, 2048)   8192        res5b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_47 (Add)                    (None, 1, 1, 2048)   0           bn5b_branch2c[0][0]              \n",
      "                                                                 activation_141[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_144 (Activation)     (None, 1, 1, 2048)   0           add_47[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2a (Conv2D)         (None, 1, 1, 512)    1049088     activation_144[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2a (BatchNormalizati (None, 1, 1, 512)    2048        res5c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_145 (Activation)     (None, 1, 1, 512)    0           bn5c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2b (Conv2D)         (None, 1, 1, 512)    2359808     activation_145[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2b (BatchNormalizati (None, 1, 1, 512)    2048        res5c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_146 (Activation)     (None, 1, 1, 512)    0           bn5c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2c (Conv2D)         (None, 1, 1, 2048)   1050624     activation_146[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2c (BatchNormalizati (None, 1, 1, 2048)   8192        res5c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_48 (Add)                    (None, 1, 1, 2048)   0           bn5c_branch2c[0][0]              \n",
      "                                                                 activation_144[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_147 (Activation)     (None, 1, 1, 2048)   0           add_48[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d_4 (Glo (None, 2048)         0           activation_147[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "dropout_3 (Dropout)             (None, 2048)         0           global_average_pooling2d_4[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 7)            14343       dropout_3[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 23,602,055\n",
      "Trainable params: 23,548,935\n",
      "Non-trainable params: 53,120\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "TODO: Image preprocessing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "8012/8012 [==============================] - 506s 63ms/step - loss: 2.0375 - accuracy: 0.5814\n",
      "Epoch 2/20\n",
      "8012/8012 [==============================] - 672s 84ms/step - loss: 1.8237 - accuracy: 0.5990\n",
      "Epoch 3/20\n",
      "8012/8012 [==============================] - 675s 84ms/step - loss: 1.5894 - accuracy: 0.6147\n",
      "Epoch 4/20\n",
      "8012/8012 [==============================] - 645s 81ms/step - loss: 1.4102 - accuracy: 0.6453\n",
      "Epoch 5/20\n",
      "8012/8012 [==============================] - 646s 81ms/step - loss: 1.2895 - accuracy: 0.6513\n",
      "Epoch 6/20\n",
      "8012/8012 [==============================] - 496s 62ms/step - loss: 1.1778 - accuracy: 0.6625\n",
      "Epoch 7/20\n",
      "8012/8012 [==============================] - 504s 63ms/step - loss: 1.1205 - accuracy: 0.6742\n",
      "Epoch 8/20\n",
      "8012/8012 [==============================] - 521s 65ms/step - loss: 1.0342 - accuracy: 0.7006\n",
      "Epoch 9/20\n",
      "8012/8012 [==============================] - 540s 67ms/step - loss: 0.9458 - accuracy: 0.7079\n",
      "Epoch 10/20\n",
      "8012/8012 [==============================] - 620s 77ms/step - loss: 0.9106 - accuracy: 0.7258\n",
      "Epoch 11/20\n",
      "8012/8012 [==============================] - 648s 81ms/step - loss: 0.8063 - accuracy: 0.7373\n",
      "Epoch 12/20\n",
      "8012/8012 [==============================] - 646s 81ms/step - loss: 0.7269 - accuracy: 0.7596\n",
      "Epoch 13/20\n",
      "8012/8012 [==============================] - 639s 80ms/step - loss: 0.7015 - accuracy: 0.7686\n",
      "Epoch 14/20\n",
      "8012/8012 [==============================] - 547s 68ms/step - loss: 0.6303 - accuracy: 0.7821\n",
      "Epoch 15/20\n",
      "8012/8012 [==============================] - 497s 62ms/step - loss: 0.5646 - accuracy: 0.8097\n",
      "Epoch 16/20\n",
      "8012/8012 [==============================] - 495s 62ms/step - loss: 0.4873 - accuracy: 0.8319\n",
      "Epoch 17/20\n",
      "8012/8012 [==============================] - 498s 62ms/step - loss: 0.4745 - accuracy: 0.8336\n",
      "Epoch 18/20\n",
      "8012/8012 [==============================] - 497s 62ms/step - loss: 0.4401 - accuracy: 0.8454\n",
      "Epoch 19/20\n",
      "8012/8012 [==============================] - 499s 62ms/step - loss: 0.4007 - accuracy: 0.8583\n",
      "Epoch 20/20\n",
      "8012/8012 [==============================] - 500s 62ms/step - loss: 0.3812 - accuracy: 0.8696\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.callbacks.History at 0x23e8ea220f0>"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(X_train, Y_train, epochs = 20, batch_size = 64)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Test model on the test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2003/2003 [==============================] - 17s 8ms/step\n",
      "Loss = 1.3395994774889364\n",
      "Test Accuracy = 0.6989516019821167\n"
     ]
    }
   ],
   "source": [
    "preds = model.evaluate(X_test, Y_test)\n",
    "print (\"Loss = \" + str(preds[0]))\n",
    "print (\"Test Accuracy = \" + str(preds[1]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Plot the confusion matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Confusion Matrix \n",
    "y_pred = model.predict(X_test)\n",
    "matrix = metrics.confusion_matrix(Y_test.argmax(axis=1), y_pred.argmax(axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 396x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#print(np.matrix(matrix))\n",
    "\n",
    "cm_df = pd.DataFrame(matrix,\n",
    "                     index = ['Melanocytic nevi','Melanoma','Benign keratosis', 'Basal cell carcinoma','Actinic keratoses','Vascular lesions', 'Dermatofibroma'], \n",
    "                     columns = ['Melanocytic nevi','Melanoma','Benign keratosis', 'Basal cell carcinoma','Actinic keratoses','Vascular lesions', 'Dermatofibroma'])\n",
    "\n",
    "plt.figure(figsize=(5.5,4))\n",
    "sns.heatmap(cm_df, annot=True)\n",
    "plt.title('Confusion Matrix'); \n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()\n",
    "\n",
    "\n",
    "#ANOTHER WAY TO DO IT\n",
    "# import seaborn as sns\n",
    "# import matplotlib.pyplot as plt     \n",
    "\n",
    "# ax= plt.subplot()\n",
    "# sns.heatmap(matrix, annot=True, ax = ax); #annot=True to annotate cells\n",
    "\n",
    "# # labels, title and ticks\n",
    "# ax.set_xlabel('Predicted labels');ax.set_ylabel('True labels'); \n",
    "# ax.set_title('Confusion Matrix'); \n",
    "# ax.xaxis.set_ticklabels(['business', 'health']); ax.yaxis.set_ticklabels(['health', 'business']);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Convolutional Neural Network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_5 (Conv2D)            (None, 32, 32, 32)        896       \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 32, 32, 32)        9248      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 16, 16, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 16, 16, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_7 (Conv2D)            (None, 16, 16, 64)        18496     \n",
      "_________________________________________________________________\n",
      "conv2d_8 (Conv2D)            (None, 16, 16, 64)        36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 8, 8, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 8, 8, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 4096)              0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 128)               524416    \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 7)                 903       \n",
      "=================================================================\n",
      "Total params: 590,887\n",
      "Trainable params: 590,887\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Set CNN model\n",
    "# Our system of layers => [[Conv2D -> relu]*2 -> MaxPool2D -> Dropout]*2 -> Flatten -> Dense -> Dropout -> Out\n",
    "input_shape = (32, 32, 3)\n",
    "num_classes = 7\n",
    "\n",
    "\n",
    "model_cnn = Sequential()\n",
    "model_cnn.add(Conv2D(32, kernel_size = (3, 3), activation = 'relu', padding = 'Same', input_shape = input_shape))\n",
    "model_cnn.add(Conv2D(32, kernel_size = (3, 3), activation = 'relu', padding = 'Same'))\n",
    "model_cnn.add(MaxPool2D(pool_size = (2, 2)))\n",
    "model_cnn.add(Dropout(0.25))\n",
    "\n",
    "model_cnn.add(Conv2D(64, kernel_size = (3, 3), activation = 'relu', padding = 'Same'))\n",
    "model_cnn.add(Conv2D(64, kernel_size = (3, 3), activation = 'relu', padding = 'Same'))\n",
    "model_cnn.add(MaxPool2D(pool_size = (2, 2)))\n",
    "model_cnn.add(Dropout(0.40))\n",
    "\n",
    "model_cnn.add(Flatten())\n",
    "model_cnn.add(Dense(128, activation='relu'))\n",
    "model_cnn.add(Dropout(0.5))\n",
    "model_cnn.add(Dense(num_classes, activation='softmax'))\n",
    "model_cnn.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define the optimizer\n",
    "optimizer = Adam(lr = .001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Compile the model\n",
    "model_cnn.compile(optimizer = optimizer, loss = 'categorical_crossentropy', metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Set learning rate annealer\n",
    "from keras.callbacks import ReduceLROnPlateau\n",
    "learning_rate_reduction = ReduceLROnPlateau(monitor = 'val_acc', patience = 3, verbose = 1, \n",
    "                                           factor = 0.5, min_lr = 0.00001)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Data Preproccessing\n",
    "datagen = ImageDataGenerator(featurewise_center = False, samplewise_center = False,\n",
    "                            featurewise_std_normalization = False, samplewise_std_normalization = False, \n",
    "                            zca_whitening = False, rotation_range = 10, zoom_range = 0.1, \n",
    "                            width_shift_range = 0.1, height_shift_range = 0.1, horizontal_flip = False, \n",
    "                            vertical_flip = False)\n",
    "datagen.fit(X_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "125/125 [==============================] - 52s 419ms/step - loss: 1.8853 - accuracy: 0.6448\n",
      "Epoch 2/20\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\GERMAN\\Anaconda3\\lib\\site-packages\\keras\\callbacks\\callbacks.py:1042: RuntimeWarning: Reduce LR on plateau conditioned on metric `val_acc` which is not available. Available metrics are: loss,accuracy,lr\n",
      "  (self.monitor, ','.join(list(logs.keys()))), RuntimeWarning\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "125/125 [==============================] - 53s 420ms/step - loss: 1.0065 - accuracy: 0.6704\n",
      "Epoch 3/20\n",
      "125/125 [==============================] - 50s 403ms/step - loss: 0.9722 - accuracy: 0.6677\n",
      "Epoch 4/20\n",
      "125/125 [==============================] - 50s 402ms/step - loss: 0.9373 - accuracy: 0.6741\n",
      "Epoch 5/20\n",
      "125/125 [==============================] - 50s 401ms/step - loss: 0.9397 - accuracy: 0.6694\n",
      "Epoch 6/20\n",
      "125/125 [==============================] - 50s 402ms/step - loss: 0.9238 - accuracy: 0.6735\n",
      "Epoch 7/20\n",
      "125/125 [==============================] - 51s 406ms/step - loss: 0.9114 - accuracy: 0.6707\n",
      "Epoch 8/20\n",
      "125/125 [==============================] - 50s 404ms/step - loss: 0.9013 - accuracy: 0.6760\n",
      "Epoch 9/20\n",
      "125/125 [==============================] - 51s 407ms/step - loss: 0.8764 - accuracy: 0.6828\n",
      "Epoch 10/20\n",
      "125/125 [==============================] - 51s 404ms/step - loss: 0.8806 - accuracy: 0.6774\n",
      "Epoch 11/20\n",
      "125/125 [==============================] - 50s 402ms/step - loss: 0.8839 - accuracy: 0.6853\n",
      "Epoch 12/20\n",
      "125/125 [==============================] - 50s 402ms/step - loss: 0.8597 - accuracy: 0.6844\n",
      "Epoch 13/20\n",
      "125/125 [==============================] - 50s 401ms/step - loss: 0.8598 - accuracy: 0.6831\n",
      "Epoch 14/20\n",
      "125/125 [==============================] - 51s 409ms/step - loss: 0.8444 - accuracy: 0.6885\n",
      "Epoch 15/20\n",
      "125/125 [==============================] - 51s 405ms/step - loss: 0.8420 - accuracy: 0.6930\n",
      "Epoch 16/20\n",
      "125/125 [==============================] - 50s 402ms/step - loss: 0.8318 - accuracy: 0.6940\n",
      "Epoch 17/20\n",
      "125/125 [==============================] - 51s 405ms/step - loss: 0.8413 - accuracy: 0.6902\n",
      "Epoch 18/20\n",
      "125/125 [==============================] - 32s 253ms/step - loss: 0.8323 - accuracy: 0.6901\n",
      "Epoch 19/20\n",
      "125/125 [==============================] - 30s 244ms/step - loss: 0.8148 - accuracy: 0.6920\n",
      "Epoch 20/20\n",
      "125/125 [==============================] - 31s 248ms/step - loss: 0.8189 - accuracy: 0.6955\n"
     ]
    }
   ],
   "source": [
    "#Training model with preprocessing and and learning rate reduction\n",
    "epochs = 20\n",
    "batch_size = 64\n",
    "history = model_cnn.fit_generator(datagen.flow(X_train, Y_train, batch_size = batch_size), \n",
    "                             epochs = epochs, verbose = 1, steps_per_epoch = X_train.shape[0] // batch_size,\n",
    "                             callbacks = [learning_rate_reduction])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "8012/8012 [==============================] - 28s 3ms/step - loss: 2.4761 - accuracy: 0.6459\n",
      "Epoch 2/20\n",
      "8012/8012 [==============================] - 28s 4ms/step - loss: 1.0290 - accuracy: 0.6670\n",
      "Epoch 3/20\n",
      "8012/8012 [==============================] - 30s 4ms/step - loss: 0.9686 - accuracy: 0.6679\n",
      "Epoch 4/20\n",
      "8012/8012 [==============================] - 30s 4ms/step - loss: 0.9414 - accuracy: 0.6696\n",
      "Epoch 5/20\n",
      "8012/8012 [==============================] - 35s 4ms/step - loss: 0.9273 - accuracy: 0.6689\n",
      "Epoch 6/20\n",
      "8012/8012 [==============================] - 32s 4ms/step - loss: 0.9046 - accuracy: 0.6735\n",
      "Epoch 7/20\n",
      "8012/8012 [==============================] - 31s 4ms/step - loss: 0.8765 - accuracy: 0.6777\n",
      "Epoch 8/20\n",
      "8012/8012 [==============================] - 36s 4ms/step - loss: 0.8681 - accuracy: 0.6826\n",
      "Epoch 9/20\n",
      "8012/8012 [==============================] - 35s 4ms/step - loss: 0.8531 - accuracy: 0.6870\n",
      "Epoch 10/20\n",
      "8012/8012 [==============================] - 32s 4ms/step - loss: 0.8332 - accuracy: 0.6926\n",
      "Epoch 11/20\n",
      "8012/8012 [==============================] - 33s 4ms/step - loss: 0.8338 - accuracy: 0.6971\n",
      "Epoch 12/20\n",
      "8012/8012 [==============================] - 32s 4ms/step - loss: 0.8238 - accuracy: 0.6968\n",
      "Epoch 13/20\n",
      "8012/8012 [==============================] - 33s 4ms/step - loss: 0.8008 - accuracy: 0.7086\n",
      "Epoch 14/20\n",
      "8012/8012 [==============================] - 30s 4ms/step - loss: 0.8005 - accuracy: 0.7059\n",
      "Epoch 15/20\n",
      "8012/8012 [==============================] - 32s 4ms/step - loss: 0.7924 - accuracy: 0.7102\n",
      "Epoch 16/20\n",
      "8012/8012 [==============================] - 31s 4ms/step - loss: 0.7718 - accuracy: 0.7132\n",
      "Epoch 17/20\n",
      "8012/8012 [==============================] - 29s 4ms/step - loss: 0.7626 - accuracy: 0.7146\n",
      "Epoch 18/20\n",
      "8012/8012 [==============================] - 31s 4ms/step - loss: 0.7639 - accuracy: 0.7168\n",
      "Epoch 19/20\n",
      "8012/8012 [==============================] - 29s 4ms/step - loss: 0.7577 - accuracy: 0.7198\n",
      "Epoch 20/20\n",
      "8012/8012 [==============================] - 29s 4ms/step - loss: 0.7555 - accuracy: 0.7195\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.callbacks.History at 0x1de8cb1af28>"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Training model without preprocessing and learning rae reduction\n",
    "model_cnn.fit(X_train, Y_train, epochs = 20, batch_size = 64)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MultiOutputClassifier(estimator=RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=10, n_jobs=1,\n",
       "            oob_score=False, random_state=1, verbose=0, warm_start=False),\n",
       "           n_jobs=-1)"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.multioutput import MultiOutputClassifier\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "forest = RandomForestClassifier(random_state=1)\n",
    "multi_target_forest = MultiOutputClassifier(forest, n_jobs=-1)\n",
    "multi_target_forest.fit(X_train, Y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 396x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Confusion Matrix \n",
    "y_pred = multi_target_forest.predict(X_test)\n",
    "matrix = metrics.confusion_matrix(Y_test.argmax(axis=1), y_pred.argmax(axis=1))\n",
    "\n",
    "cm_df = pd.DataFrame(matrix,\n",
    "                     index = ['Melanocytic nevi','Melanoma','Benign keratosis', 'Basal cell carcinoma','Actinic keratoses','Vascular lesions', 'Dermatofibroma'], \n",
    "                     columns = ['Melanocytic nevi','Melanoma','Benign keratosis', 'Basal cell carcinoma','Actinic keratoses','Vascular lesions', 'Dermatofibroma'])\n",
    "\n",
    "plt.figure(figsize=(5.5,4))\n",
    "sns.heatmap(cm_df, annot=True)\n",
    "plt.title('Confusion Matrix'); \n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SVM"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "OneVsRestClassifier(estimator=LinearSVC(C=1.0, class_weight=None, dual=True, fit_intercept=True,\n",
       "     intercept_scaling=1, loss='squared_hinge', max_iter=1000,\n",
       "     multi_class='ovr', penalty='l2', random_state=None, tol=0.0001,\n",
       "     verbose=0),\n",
       "          n_jobs=1)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.multiclass import OneVsRestClassifier\n",
    "from sklearn.svm import LinearSVC\n",
    "\n",
    "clf = OneVsRestClassifier(LinearSVC())\n",
    "clf.fit(X_train, Y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 396x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Confusion Matrix \n",
    "y_pred = clf.predict(X_test)\n",
    "matrix = metrics.confusion_matrix(Y_test.argmax(axis=1), y_pred.argmax(axis=1))\n",
    "\n",
    "cm_df = pd.DataFrame(matrix,\n",
    "                     index = ['Melanocytic nevi','Melanoma','Benign keratosis', 'Basal cell carcinoma','Actinic keratoses','Vascular lesions', 'Dermatofibroma'], \n",
    "                     columns = ['Melanocytic nevi','Melanoma','Benign keratosis', 'Basal cell carcinoma','Actinic keratoses','Vascular lesions', 'Dermatofibroma'])\n",
    "\n",
    "plt.figure(figsize=(5.5,4))\n",
    "sns.heatmap(cm_df, annot=True)\n",
    "plt.title('Confusion Matrix'); \n",
    "plt.ylabel('True label')\n",
    "plt.xlabel('Predicted label')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "TODO: Analysis of accuracies for various classifiers"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}