SkinPigmentNotebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
    "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:37.730168Z",
     "iopub.status.busy": "2021-02-03T14:59:37.729423Z",
     "iopub.status.idle": "2021-02-03T14:59:45.581792Z",
     "shell.execute_reply": "2021-02-03T14:59:45.582361Z"
    },
    "papermill": {
     "duration": 7.873903,
     "end_time": "2021-02-03T14:59:45.582710",
     "exception": false,
     "start_time": "2021-02-03T14:59:37.708807",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import tensorflow as tf\n",
    "\n",
    "from sklearn.metrics import confusion_matrix, classification_report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:45.612834Z",
     "iopub.status.busy": "2021-02-03T14:59:45.611846Z",
     "iopub.status.idle": "2021-02-03T14:59:47.469082Z",
     "shell.execute_reply": "2021-02-03T14:59:47.469674Z"
    },
    "papermill": {
     "duration": 1.873801,
     "end_time": "2021-02-03T14:59:47.469882",
     "exception": false,
     "start_time": "2021-02-03T14:59:45.596081",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "data = pd.read_csv('../input/skin-cancer-mnist-ham10000/hmnist_28_28_L.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.500244Z",
     "iopub.status.busy": "2021-02-03T14:59:47.499178Z",
     "iopub.status.idle": "2021-02-03T14:59:47.545132Z",
     "shell.execute_reply": "2021-02-03T14:59:47.545665Z"
    },
    "papermill": {
     "duration": 0.062963,
     "end_time": "2021-02-03T14:59:47.545845",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.482882",
     "status": "completed"
    },
    "tags": []
   },
   "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>pixel0775</th>\n",
       "      <th>pixel0776</th>\n",
       "      <th>pixel0777</th>\n",
       "      <th>pixel0778</th>\n",
       "      <th>pixel0779</th>\n",
       "      <th>pixel0780</th>\n",
       "      <th>pixel0781</th>\n",
       "      <th>pixel0782</th>\n",
       "      <th>pixel0783</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>169</td>\n",
       "      <td>171</td>\n",
       "      <td>170</td>\n",
       "      <td>177</td>\n",
       "      <td>181</td>\n",
       "      <td>182</td>\n",
       "      <td>181</td>\n",
       "      <td>185</td>\n",
       "      <td>194</td>\n",
       "      <td>192</td>\n",
       "      <td>...</td>\n",
       "      <td>184</td>\n",
       "      <td>186</td>\n",
       "      <td>185</td>\n",
       "      <td>180</td>\n",
       "      <td>157</td>\n",
       "      <td>140</td>\n",
       "      <td>140</td>\n",
       "      <td>159</td>\n",
       "      <td>165</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>19</td>\n",
       "      <td>57</td>\n",
       "      <td>105</td>\n",
       "      <td>140</td>\n",
       "      <td>149</td>\n",
       "      <td>148</td>\n",
       "      <td>144</td>\n",
       "      <td>155</td>\n",
       "      <td>170</td>\n",
       "      <td>170</td>\n",
       "      <td>...</td>\n",
       "      <td>172</td>\n",
       "      <td>175</td>\n",
       "      <td>160</td>\n",
       "      <td>144</td>\n",
       "      <td>114</td>\n",
       "      <td>89</td>\n",
       "      <td>47</td>\n",
       "      <td>18</td>\n",
       "      <td>18</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>155</td>\n",
       "      <td>163</td>\n",
       "      <td>161</td>\n",
       "      <td>167</td>\n",
       "      <td>167</td>\n",
       "      <td>172</td>\n",
       "      <td>155</td>\n",
       "      <td>152</td>\n",
       "      <td>165</td>\n",
       "      <td>175</td>\n",
       "      <td>...</td>\n",
       "      <td>163</td>\n",
       "      <td>178</td>\n",
       "      <td>157</td>\n",
       "      <td>166</td>\n",
       "      <td>167</td>\n",
       "      <td>148</td>\n",
       "      <td>141</td>\n",
       "      <td>136</td>\n",
       "      <td>115</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>25</td>\n",
       "      <td>71</td>\n",
       "      <td>116</td>\n",
       "      <td>139</td>\n",
       "      <td>136</td>\n",
       "      <td>153</td>\n",
       "      <td>148</td>\n",
       "      <td>161</td>\n",
       "      <td>172</td>\n",
       "      <td>162</td>\n",
       "      <td>...</td>\n",
       "      <td>125</td>\n",
       "      <td>135</td>\n",
       "      <td>138</td>\n",
       "      <td>137</td>\n",
       "      <td>111</td>\n",
       "      <td>71</td>\n",
       "      <td>32</td>\n",
       "      <td>16</td>\n",
       "      <td>16</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>129</td>\n",
       "      <td>162</td>\n",
       "      <td>181</td>\n",
       "      <td>196</td>\n",
       "      <td>205</td>\n",
       "      <td>208</td>\n",
       "      <td>205</td>\n",
       "      <td>213</td>\n",
       "      <td>225</td>\n",
       "      <td>224</td>\n",
       "      <td>...</td>\n",
       "      <td>210</td>\n",
       "      <td>197</td>\n",
       "      <td>172</td>\n",
       "      <td>190</td>\n",
       "      <td>195</td>\n",
       "      <td>193</td>\n",
       "      <td>181</td>\n",
       "      <td>147</td>\n",
       "      <td>88</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10010</th>\n",
       "      <td>172</td>\n",
       "      <td>171</td>\n",
       "      <td>173</td>\n",
       "      <td>175</td>\n",
       "      <td>164</td>\n",
       "      <td>187</td>\n",
       "      <td>207</td>\n",
       "      <td>210</td>\n",
       "      <td>208</td>\n",
       "      <td>206</td>\n",
       "      <td>...</td>\n",
       "      <td>210</td>\n",
       "      <td>217</td>\n",
       "      <td>221</td>\n",
       "      <td>209</td>\n",
       "      <td>185</td>\n",
       "      <td>187</td>\n",
       "      <td>192</td>\n",
       "      <td>192</td>\n",
       "      <td>192</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10011</th>\n",
       "      <td>2</td>\n",
       "      <td>34</td>\n",
       "      <td>108</td>\n",
       "      <td>116</td>\n",
       "      <td>114</td>\n",
       "      <td>119</td>\n",
       "      <td>131</td>\n",
       "      <td>139</td>\n",
       "      <td>139</td>\n",
       "      <td>145</td>\n",
       "      <td>...</td>\n",
       "      <td>173</td>\n",
       "      <td>169</td>\n",
       "      <td>168</td>\n",
       "      <td>168</td>\n",
       "      <td>143</td>\n",
       "      <td>138</td>\n",
       "      <td>83</td>\n",
       "      <td>23</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10012</th>\n",
       "      <td>122</td>\n",
       "      <td>154</td>\n",
       "      <td>162</td>\n",
       "      <td>170</td>\n",
       "      <td>179</td>\n",
       "      <td>197</td>\n",
       "      <td>200</td>\n",
       "      <td>195</td>\n",
       "      <td>202</td>\n",
       "      <td>199</td>\n",
       "      <td>...</td>\n",
       "      <td>221</td>\n",
       "      <td>215</td>\n",
       "      <td>205</td>\n",
       "      <td>187</td>\n",
       "      <td>209</td>\n",
       "      <td>198</td>\n",
       "      <td>187</td>\n",
       "      <td>164</td>\n",
       "      <td>156</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10013</th>\n",
       "      <td>137</td>\n",
       "      <td>143</td>\n",
       "      <td>141</td>\n",
       "      <td>139</td>\n",
       "      <td>147</td>\n",
       "      <td>152</td>\n",
       "      <td>155</td>\n",
       "      <td>152</td>\n",
       "      <td>155</td>\n",
       "      <td>159</td>\n",
       "      <td>...</td>\n",
       "      <td>172</td>\n",
       "      <td>171</td>\n",
       "      <td>175</td>\n",
       "      <td>183</td>\n",
       "      <td>177</td>\n",
       "      <td>170</td>\n",
       "      <td>169</td>\n",
       "      <td>166</td>\n",
       "      <td>170</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10014</th>\n",
       "      <td>149</td>\n",
       "      <td>157</td>\n",
       "      <td>157</td>\n",
       "      <td>152</td>\n",
       "      <td>151</td>\n",
       "      <td>152</td>\n",
       "      <td>154</td>\n",
       "      <td>163</td>\n",
       "      <td>149</td>\n",
       "      <td>145</td>\n",
       "      <td>...</td>\n",
       "      <td>159</td>\n",
       "      <td>154</td>\n",
       "      <td>145</td>\n",
       "      <td>106</td>\n",
       "      <td>67</td>\n",
       "      <td>62</td>\n",
       "      <td>103</td>\n",
       "      <td>146</td>\n",
       "      <td>148</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10015 rows × 785 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       pixel0000  pixel0001  pixel0002  pixel0003  pixel0004  pixel0005  \\\n",
       "0            169        171        170        177        181        182   \n",
       "1             19         57        105        140        149        148   \n",
       "2            155        163        161        167        167        172   \n",
       "3             25         71        116        139        136        153   \n",
       "4            129        162        181        196        205        208   \n",
       "...          ...        ...        ...        ...        ...        ...   \n",
       "10010        172        171        173        175        164        187   \n",
       "10011          2         34        108        116        114        119   \n",
       "10012        122        154        162        170        179        197   \n",
       "10013        137        143        141        139        147        152   \n",
       "10014        149        157        157        152        151        152   \n",
       "\n",
       "       pixel0006  pixel0007  pixel0008  pixel0009  ...  pixel0775  pixel0776  \\\n",
       "0            181        185        194        192  ...        184        186   \n",
       "1            144        155        170        170  ...        172        175   \n",
       "2            155        152        165        175  ...        163        178   \n",
       "3            148        161        172        162  ...        125        135   \n",
       "4            205        213        225        224  ...        210        197   \n",
       "...          ...        ...        ...        ...  ...        ...        ...   \n",
       "10010        207        210        208        206  ...        210        217   \n",
       "10011        131        139        139        145  ...        173        169   \n",
       "10012        200        195        202        199  ...        221        215   \n",
       "10013        155        152        155        159  ...        172        171   \n",
       "10014        154        163        149        145  ...        159        154   \n",
       "\n",
       "       pixel0777  pixel0778  pixel0779  pixel0780  pixel0781  pixel0782  \\\n",
       "0            185        180        157        140        140        159   \n",
       "1            160        144        114         89         47         18   \n",
       "2            157        166        167        148        141        136   \n",
       "3            138        137        111         71         32         16   \n",
       "4            172        190        195        193        181        147   \n",
       "...          ...        ...        ...        ...        ...        ...   \n",
       "10010        221        209        185        187        192        192   \n",
       "10011        168        168        143        138         83         23   \n",
       "10012        205        187        209        198        187        164   \n",
       "10013        175        183        177        170        169        166   \n",
       "10014        145        106         67         62        103        146   \n",
       "\n",
       "       pixel0783  label  \n",
       "0            165      2  \n",
       "1             18      2  \n",
       "2            115      2  \n",
       "3             16      2  \n",
       "4             88      2  \n",
       "...          ...    ...  \n",
       "10010        192      0  \n",
       "10011          3      0  \n",
       "10012        156      0  \n",
       "10013        170      0  \n",
       "10014        148      6  \n",
       "\n",
       "[10015 rows x 785 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.578521Z",
     "iopub.status.busy": "2021-02-03T14:59:47.577371Z",
     "iopub.status.idle": "2021-02-03T14:59:47.597469Z",
     "shell.execute_reply": "2021-02-03T14:59:47.597947Z"
    },
    "papermill": {
     "duration": 0.037853,
     "end_time": "2021-02-03T14:59:47.598139",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.560286",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.isna().sum().sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.630909Z",
     "iopub.status.busy": "2021-02-03T14:59:47.629947Z",
     "iopub.status.idle": "2021-02-03T14:59:47.692137Z",
     "shell.execute_reply": "2021-02-03T14:59:47.692655Z"
    },
    "papermill": {
     "duration": 0.080111,
     "end_time": "2021-02-03T14:59:47.692852",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.612741",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "y = data['label'].copy()\n",
    "X = data.drop('label', axis=1).copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.726244Z",
     "iopub.status.busy": "2021-02-03T14:59:47.725371Z",
     "iopub.status.idle": "2021-02-03T14:59:47.736690Z",
     "shell.execute_reply": "2021-02-03T14:59:47.736100Z"
    },
    "papermill": {
     "duration": 0.02927,
     "end_time": "2021-02-03T14:59:47.736836",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.707566",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4    6705\n",
       "6    1113\n",
       "2    1099\n",
       "1     514\n",
       "0     327\n",
       "5     142\n",
       "3     115\n",
       "Name: label, dtype: int64"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.773696Z",
     "iopub.status.busy": "2021-02-03T14:59:47.772406Z",
     "iopub.status.idle": "2021-02-03T14:59:47.775832Z",
     "shell.execute_reply": "2021-02-03T14:59:47.775157Z"
    },
    "papermill": {
     "duration": 0.023895,
     "end_time": "2021-02-03T14:59:47.775980",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.752085",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "label_mapping = {\n",
    "    0: 'nv',\n",
    "    1: 'mel',\n",
    "    2: 'bkl',\n",
    "    3: 'bcc',\n",
    "    4: 'akiec',\n",
    "    5: 'vasc',\n",
    "    6: 'df'\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.811758Z",
     "iopub.status.busy": "2021-02-03T14:59:47.810822Z",
     "iopub.status.idle": "2021-02-03T14:59:47.883137Z",
     "shell.execute_reply": "2021-02-03T14:59:47.883681Z"
    },
    "papermill": {
     "duration": 0.092439,
     "end_time": "2021-02-03T14:59:47.883859",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.791420",
     "status": "completed"
    },
    "tags": []
   },
   "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>pixel0774</th>\n",
       "      <th>pixel0775</th>\n",
       "      <th>pixel0776</th>\n",
       "      <th>pixel0777</th>\n",
       "      <th>pixel0778</th>\n",
       "      <th>pixel0779</th>\n",
       "      <th>pixel0780</th>\n",
       "      <th>pixel0781</th>\n",
       "      <th>pixel0782</th>\n",
       "      <th>pixel0783</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.662745</td>\n",
       "      <td>0.670588</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.694118</td>\n",
       "      <td>0.709804</td>\n",
       "      <td>0.713725</td>\n",
       "      <td>0.709804</td>\n",
       "      <td>0.725490</td>\n",
       "      <td>0.760784</td>\n",
       "      <td>0.752941</td>\n",
       "      <td>...</td>\n",
       "      <td>0.690196</td>\n",
       "      <td>0.721569</td>\n",
       "      <td>0.729412</td>\n",
       "      <td>0.725490</td>\n",
       "      <td>0.705882</td>\n",
       "      <td>0.615686</td>\n",
       "      <td>0.549020</td>\n",
       "      <td>0.549020</td>\n",
       "      <td>0.623529</td>\n",
       "      <td>0.647059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.074510</td>\n",
       "      <td>0.223529</td>\n",
       "      <td>0.411765</td>\n",
       "      <td>0.549020</td>\n",
       "      <td>0.584314</td>\n",
       "      <td>0.580392</td>\n",
       "      <td>0.564706</td>\n",
       "      <td>0.607843</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>...</td>\n",
       "      <td>0.709804</td>\n",
       "      <td>0.674510</td>\n",
       "      <td>0.686275</td>\n",
       "      <td>0.627451</td>\n",
       "      <td>0.564706</td>\n",
       "      <td>0.447059</td>\n",
       "      <td>0.349020</td>\n",
       "      <td>0.184314</td>\n",
       "      <td>0.070588</td>\n",
       "      <td>0.070588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.607843</td>\n",
       "      <td>0.639216</td>\n",
       "      <td>0.631373</td>\n",
       "      <td>0.654902</td>\n",
       "      <td>0.654902</td>\n",
       "      <td>0.674510</td>\n",
       "      <td>0.607843</td>\n",
       "      <td>0.596078</td>\n",
       "      <td>0.647059</td>\n",
       "      <td>0.686275</td>\n",
       "      <td>...</td>\n",
       "      <td>0.635294</td>\n",
       "      <td>0.639216</td>\n",
       "      <td>0.698039</td>\n",
       "      <td>0.615686</td>\n",
       "      <td>0.650980</td>\n",
       "      <td>0.654902</td>\n",
       "      <td>0.580392</td>\n",
       "      <td>0.552941</td>\n",
       "      <td>0.533333</td>\n",
       "      <td>0.450980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.098039</td>\n",
       "      <td>0.278431</td>\n",
       "      <td>0.454902</td>\n",
       "      <td>0.545098</td>\n",
       "      <td>0.533333</td>\n",
       "      <td>0.600000</td>\n",
       "      <td>0.580392</td>\n",
       "      <td>0.631373</td>\n",
       "      <td>0.674510</td>\n",
       "      <td>0.635294</td>\n",
       "      <td>...</td>\n",
       "      <td>0.556863</td>\n",
       "      <td>0.490196</td>\n",
       "      <td>0.529412</td>\n",
       "      <td>0.541176</td>\n",
       "      <td>0.537255</td>\n",
       "      <td>0.435294</td>\n",
       "      <td>0.278431</td>\n",
       "      <td>0.125490</td>\n",
       "      <td>0.062745</td>\n",
       "      <td>0.062745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.505882</td>\n",
       "      <td>0.635294</td>\n",
       "      <td>0.709804</td>\n",
       "      <td>0.768627</td>\n",
       "      <td>0.803922</td>\n",
       "      <td>0.815686</td>\n",
       "      <td>0.803922</td>\n",
       "      <td>0.835294</td>\n",
       "      <td>0.882353</td>\n",
       "      <td>0.878431</td>\n",
       "      <td>...</td>\n",
       "      <td>0.827451</td>\n",
       "      <td>0.823529</td>\n",
       "      <td>0.772549</td>\n",
       "      <td>0.674510</td>\n",
       "      <td>0.745098</td>\n",
       "      <td>0.764706</td>\n",
       "      <td>0.756863</td>\n",
       "      <td>0.709804</td>\n",
       "      <td>0.576471</td>\n",
       "      <td>0.345098</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10010</th>\n",
       "      <td>0.674510</td>\n",
       "      <td>0.670588</td>\n",
       "      <td>0.678431</td>\n",
       "      <td>0.686275</td>\n",
       "      <td>0.643137</td>\n",
       "      <td>0.733333</td>\n",
       "      <td>0.811765</td>\n",
       "      <td>0.823529</td>\n",
       "      <td>0.815686</td>\n",
       "      <td>0.807843</td>\n",
       "      <td>...</td>\n",
       "      <td>0.823529</td>\n",
       "      <td>0.823529</td>\n",
       "      <td>0.850980</td>\n",
       "      <td>0.866667</td>\n",
       "      <td>0.819608</td>\n",
       "      <td>0.725490</td>\n",
       "      <td>0.733333</td>\n",
       "      <td>0.752941</td>\n",
       "      <td>0.752941</td>\n",
       "      <td>0.752941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10011</th>\n",
       "      <td>0.007843</td>\n",
       "      <td>0.133333</td>\n",
       "      <td>0.423529</td>\n",
       "      <td>0.454902</td>\n",
       "      <td>0.447059</td>\n",
       "      <td>0.466667</td>\n",
       "      <td>0.513725</td>\n",
       "      <td>0.545098</td>\n",
       "      <td>0.545098</td>\n",
       "      <td>0.568627</td>\n",
       "      <td>...</td>\n",
       "      <td>0.721569</td>\n",
       "      <td>0.678431</td>\n",
       "      <td>0.662745</td>\n",
       "      <td>0.658824</td>\n",
       "      <td>0.658824</td>\n",
       "      <td>0.560784</td>\n",
       "      <td>0.541176</td>\n",
       "      <td>0.325490</td>\n",
       "      <td>0.090196</td>\n",
       "      <td>0.011765</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10012</th>\n",
       "      <td>0.478431</td>\n",
       "      <td>0.603922</td>\n",
       "      <td>0.635294</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.701961</td>\n",
       "      <td>0.772549</td>\n",
       "      <td>0.784314</td>\n",
       "      <td>0.764706</td>\n",
       "      <td>0.792157</td>\n",
       "      <td>0.780392</td>\n",
       "      <td>...</td>\n",
       "      <td>0.874510</td>\n",
       "      <td>0.866667</td>\n",
       "      <td>0.843137</td>\n",
       "      <td>0.803922</td>\n",
       "      <td>0.733333</td>\n",
       "      <td>0.819608</td>\n",
       "      <td>0.776471</td>\n",
       "      <td>0.733333</td>\n",
       "      <td>0.643137</td>\n",
       "      <td>0.611765</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10013</th>\n",
       "      <td>0.537255</td>\n",
       "      <td>0.560784</td>\n",
       "      <td>0.552941</td>\n",
       "      <td>0.545098</td>\n",
       "      <td>0.576471</td>\n",
       "      <td>0.596078</td>\n",
       "      <td>0.607843</td>\n",
       "      <td>0.596078</td>\n",
       "      <td>0.607843</td>\n",
       "      <td>0.623529</td>\n",
       "      <td>...</td>\n",
       "      <td>0.698039</td>\n",
       "      <td>0.674510</td>\n",
       "      <td>0.670588</td>\n",
       "      <td>0.686275</td>\n",
       "      <td>0.717647</td>\n",
       "      <td>0.694118</td>\n",
       "      <td>0.666667</td>\n",
       "      <td>0.662745</td>\n",
       "      <td>0.650980</td>\n",
       "      <td>0.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10014</th>\n",
       "      <td>0.584314</td>\n",
       "      <td>0.615686</td>\n",
       "      <td>0.615686</td>\n",
       "      <td>0.596078</td>\n",
       "      <td>0.592157</td>\n",
       "      <td>0.596078</td>\n",
       "      <td>0.603922</td>\n",
       "      <td>0.639216</td>\n",
       "      <td>0.584314</td>\n",
       "      <td>0.568627</td>\n",
       "      <td>...</td>\n",
       "      <td>0.650980</td>\n",
       "      <td>0.623529</td>\n",
       "      <td>0.603922</td>\n",
       "      <td>0.568627</td>\n",
       "      <td>0.415686</td>\n",
       "      <td>0.262745</td>\n",
       "      <td>0.243137</td>\n",
       "      <td>0.403922</td>\n",
       "      <td>0.572549</td>\n",
       "      <td>0.580392</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10015 rows × 784 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       pixel0000  pixel0001  pixel0002  pixel0003  pixel0004  pixel0005  \\\n",
       "0       0.662745   0.670588   0.666667   0.694118   0.709804   0.713725   \n",
       "1       0.074510   0.223529   0.411765   0.549020   0.584314   0.580392   \n",
       "2       0.607843   0.639216   0.631373   0.654902   0.654902   0.674510   \n",
       "3       0.098039   0.278431   0.454902   0.545098   0.533333   0.600000   \n",
       "4       0.505882   0.635294   0.709804   0.768627   0.803922   0.815686   \n",
       "...          ...        ...        ...        ...        ...        ...   \n",
       "10010   0.674510   0.670588   0.678431   0.686275   0.643137   0.733333   \n",
       "10011   0.007843   0.133333   0.423529   0.454902   0.447059   0.466667   \n",
       "10012   0.478431   0.603922   0.635294   0.666667   0.701961   0.772549   \n",
       "10013   0.537255   0.560784   0.552941   0.545098   0.576471   0.596078   \n",
       "10014   0.584314   0.615686   0.615686   0.596078   0.592157   0.596078   \n",
       "\n",
       "       pixel0006  pixel0007  pixel0008  pixel0009  ...  pixel0774  pixel0775  \\\n",
       "0       0.709804   0.725490   0.760784   0.752941  ...   0.690196   0.721569   \n",
       "1       0.564706   0.607843   0.666667   0.666667  ...   0.709804   0.674510   \n",
       "2       0.607843   0.596078   0.647059   0.686275  ...   0.635294   0.639216   \n",
       "3       0.580392   0.631373   0.674510   0.635294  ...   0.556863   0.490196   \n",
       "4       0.803922   0.835294   0.882353   0.878431  ...   0.827451   0.823529   \n",
       "...          ...        ...        ...        ...  ...        ...        ...   \n",
       "10010   0.811765   0.823529   0.815686   0.807843  ...   0.823529   0.823529   \n",
       "10011   0.513725   0.545098   0.545098   0.568627  ...   0.721569   0.678431   \n",
       "10012   0.784314   0.764706   0.792157   0.780392  ...   0.874510   0.866667   \n",
       "10013   0.607843   0.596078   0.607843   0.623529  ...   0.698039   0.674510   \n",
       "10014   0.603922   0.639216   0.584314   0.568627  ...   0.650980   0.623529   \n",
       "\n",
       "       pixel0776  pixel0777  pixel0778  pixel0779  pixel0780  pixel0781  \\\n",
       "0       0.729412   0.725490   0.705882   0.615686   0.549020   0.549020   \n",
       "1       0.686275   0.627451   0.564706   0.447059   0.349020   0.184314   \n",
       "2       0.698039   0.615686   0.650980   0.654902   0.580392   0.552941   \n",
       "3       0.529412   0.541176   0.537255   0.435294   0.278431   0.125490   \n",
       "4       0.772549   0.674510   0.745098   0.764706   0.756863   0.709804   \n",
       "...          ...        ...        ...        ...        ...        ...   \n",
       "10010   0.850980   0.866667   0.819608   0.725490   0.733333   0.752941   \n",
       "10011   0.662745   0.658824   0.658824   0.560784   0.541176   0.325490   \n",
       "10012   0.843137   0.803922   0.733333   0.819608   0.776471   0.733333   \n",
       "10013   0.670588   0.686275   0.717647   0.694118   0.666667   0.662745   \n",
       "10014   0.603922   0.568627   0.415686   0.262745   0.243137   0.403922   \n",
       "\n",
       "       pixel0782  pixel0783  \n",
       "0       0.623529   0.647059  \n",
       "1       0.070588   0.070588  \n",
       "2       0.533333   0.450980  \n",
       "3       0.062745   0.062745  \n",
       "4       0.576471   0.345098  \n",
       "...          ...        ...  \n",
       "10010   0.752941   0.752941  \n",
       "10011   0.090196   0.011765  \n",
       "10012   0.643137   0.611765  \n",
       "10013   0.650980   0.666667  \n",
       "10014   0.572549   0.580392  \n",
       "\n",
       "[10015 rows x 784 columns]"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X = X / 255\n",
    "\n",
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.922003Z",
     "iopub.status.busy": "2021-02-03T14:59:47.921216Z",
     "iopub.status.idle": "2021-02-03T14:59:47.949720Z",
     "shell.execute_reply": "2021-02-03T14:59:47.949074Z"
    },
    "papermill": {
     "duration": 0.04949,
     "end_time": "2021-02-03T14:59:47.949892",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.900402",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10015, 784)\n"
     ]
    }
   ],
   "source": [
    "X = np.array(X)\n",
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:47.991658Z",
     "iopub.status.busy": "2021-02-03T14:59:47.990812Z",
     "iopub.status.idle": "2021-02-03T14:59:47.994475Z",
     "shell.execute_reply": "2021-02-03T14:59:47.995005Z"
    },
    "papermill": {
     "duration": 0.027741,
     "end_time": "2021-02-03T14:59:47.995180",
     "exception": false,
     "start_time": "2021-02-03T14:59:47.967439",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10015, 28, 28, 1)\n"
     ]
    }
   ],
   "source": [
    "X = X.reshape(-1, 28, 28, 1)\n",
    "print(X.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:48.055305Z",
     "iopub.status.busy": "2021-02-03T14:59:48.042394Z",
     "iopub.status.idle": "2021-02-03T14:59:48.060304Z",
     "shell.execute_reply": "2021-02-03T14:59:48.059668Z"
    },
    "papermill": {
     "duration": 0.046939,
     "end_time": "2021-02-03T14:59:48.060477",
     "exception": false,
     "start_time": "2021-02-03T14:59:48.013538",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "sample_data = pd.Series(list(zip(X, y))).sample(9)\n",
    "\n",
    "sample_X = np.stack(np.array(sample_data.apply(lambda x: x[0])))\n",
    "sample_y = np.array(sample_data.apply(lambda x: x[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:48.118344Z",
     "iopub.status.busy": "2021-02-03T14:59:48.115851Z",
     "iopub.status.idle": "2021-02-03T14:59:48.619378Z",
     "shell.execute_reply": "2021-02-03T14:59:48.618638Z"
    },
    "papermill": {
     "duration": 0.541651,
     "end_time": "2021-02-03T14:59:48.619537",
     "exception": false,
     "start_time": "2021-02-03T14:59:48.077886",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x864 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12, 12))\n",
    "\n",
    "for i in range(9):\n",
    "    plt.subplot(3, 3, i + 1)\n",
    "    plt.imshow(np.squeeze(sample_X[i]))\n",
    "    img_label = label_mapping[sample_y[i]]\n",
    "    plt.title(img_label)\n",
    "    plt.axis(\"off\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:48.665276Z",
     "iopub.status.busy": "2021-02-03T14:59:48.664581Z",
     "iopub.status.idle": "2021-02-03T14:59:48.787128Z",
     "shell.execute_reply": "2021-02-03T14:59:48.786284Z"
    },
    "papermill": {
     "duration": 0.147943,
     "end_time": "2021-02-03T14:59:48.787303",
     "exception": false,
     "start_time": "2021-02-03T14:59:48.639360",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = train_test_split(X, y, train_size=0.7, random_state=123)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:48.835567Z",
     "iopub.status.busy": "2021-02-03T14:59:48.834874Z",
     "iopub.status.idle": "2021-02-03T14:59:48.979978Z",
     "shell.execute_reply": "2021-02-03T14:59:48.979031Z"
    },
    "papermill": {
     "duration": 0.173761,
     "end_time": "2021-02-03T14:59:48.980201",
     "exception": false,
     "start_time": "2021-02-03T14:59:48.806440",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input (InputLayer)           [(None, 28, 28, 1)]       0         \n",
      "_________________________________________________________________\n",
      "conv1 (Conv2D)               (None, 26, 26, 16)        160       \n",
      "_________________________________________________________________\n",
      "maxpool1 (MaxPooling2D)      (None, 13, 13, 16)        0         \n",
      "_________________________________________________________________\n",
      "conv2 (Conv2D)               (None, 11, 11, 32)        4640      \n",
      "_________________________________________________________________\n",
      "maxpool2 (MaxPooling2D)      (None, 5, 5, 32)          0         \n",
      "_________________________________________________________________\n",
      "conv3 (Conv2D)               (None, 3, 3, 64)          18496     \n",
      "_________________________________________________________________\n",
      "maxpool3 (MaxPooling2D)      (None, 1, 1, 64)          0         \n",
      "_________________________________________________________________\n",
      "avgpool (GlobalAveragePoolin (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "output (Dense)               (None, 7)                 455       \n",
      "=================================================================\n",
      "Total params: 23,751\n",
      "Trainable params: 23,751\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n",
      "('Failed to import pydot. You must `pip install pydot` and install graphviz (https://graphviz.gitlab.io/download/), ', 'for `pydotprint` to work.')\n"
     ]
    }
   ],
   "source": [
    "\n",
    "inputs = tf.keras.Input(shape=(28, 28, 1), name='input')\n",
    "\n",
    "conv1 = tf.keras.layers.Conv2D(filters=16, kernel_size=3, name='conv1')(inputs)\n",
    "maxpool1 = tf.keras.layers.MaxPooling2D(name='maxpool1')(conv1)\n",
    "\n",
    "conv2 = tf.keras.layers.Conv2D(filters=32, kernel_size=3, name='conv2')(maxpool1)\n",
    "maxpool2 = tf.keras.layers.MaxPooling2D(name='maxpool2')(conv2)\n",
    "\n",
    "conv3 = tf.keras.layers.Conv2D(filters=64, kernel_size=3, name='conv3')(maxpool2)\n",
    "maxpool3 = tf.keras.layers.MaxPooling2D(name='maxpool3')(conv3)\n",
    "\n",
    "avgpool = tf.keras.layers.GlobalAveragePooling2D(name='avgpool')(maxpool3)\n",
    "\n",
    "outputs = tf.keras.layers.Dense(7, activation='softmax', name='output')(avgpool)\n",
    "\n",
    "\n",
    "model = tf.keras.Model(inputs=inputs, outputs=outputs)\n",
    "\n",
    "print(model.summary())\n",
    "tf.keras.utils.plot_model(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T14:59:49.039381Z",
     "iopub.status.busy": "2021-02-03T14:59:49.034956Z",
     "iopub.status.idle": "2021-02-03T15:00:10.998761Z",
     "shell.execute_reply": "2021-02-03T15:00:10.999276Z"
    },
    "papermill": {
     "duration": 21.998645,
     "end_time": "2021-02-03T15:00:10.999474",
     "exception": false,
     "start_time": "2021-02-03T14:59:49.000829",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "176/176 [==============================] - 3s 15ms/step - loss: 1.2040 - accuracy: 0.6698 - val_loss: 1.0990 - val_accuracy: 0.6583\n",
      "Epoch 2/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 1.0517 - accuracy: 0.6695 - val_loss: 1.0320 - val_accuracy: 0.6591\n",
      "Epoch 3/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.9907 - accuracy: 0.6791 - val_loss: 1.0013 - val_accuracy: 0.6605\n",
      "Epoch 4/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.9625 - accuracy: 0.6819 - val_loss: 0.9856 - val_accuracy: 0.6633\n",
      "Epoch 5/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.9414 - accuracy: 0.6890 - val_loss: 0.9527 - val_accuracy: 0.6662\n",
      "Epoch 6/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.8989 - accuracy: 0.6929 - val_loss: 0.9462 - val_accuracy: 0.6683\n",
      "Epoch 7/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.8926 - accuracy: 0.6830 - val_loss: 0.9356 - val_accuracy: 0.6683\n",
      "Epoch 8/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.8698 - accuracy: 0.6946 - val_loss: 0.9512 - val_accuracy: 0.6733\n",
      "Epoch 9/50\n",
      "176/176 [==============================] - 2s 12ms/step - loss: 0.8804 - accuracy: 0.6906 - val_loss: 0.9551 - val_accuracy: 0.6826\n",
      "Epoch 10/50\n",
      "176/176 [==============================] - 2s 11ms/step - loss: 0.8640 - accuracy: 0.7017 - val_loss: 0.9866 - val_accuracy: 0.6583\n"
     ]
    }
   ],
   "source": [
    "model.compile(\n",
    "    optimizer='adam',\n",
    "    loss='sparse_categorical_crossentropy',\n",
    "    metrics=['accuracy']\n",
    ")\n",
    "\n",
    "\n",
    "batch_size = 32\n",
    "epochs = 50\n",
    "\n",
    "history = model.fit(\n",
    "    X_train,\n",
    "    y_train,\n",
    "    validation_split=0.2,\n",
    "    batch_size=batch_size,\n",
    "    epochs=epochs,\n",
    "    callbacks=[\n",
    "        tf.keras.callbacks.EarlyStopping(\n",
    "            monitor='val_loss',\n",
    "            patience=3,\n",
    "            restore_best_weights=True\n",
    "        )\n",
    "    ]\n",
    ")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T15:00:11.257518Z",
     "iopub.status.busy": "2021-02-03T15:00:11.256797Z",
     "iopub.status.idle": "2021-02-03T15:00:11.645277Z",
     "shell.execute_reply": "2021-02-03T15:00:11.644100Z"
    },
    "papermill": {
     "duration": 0.520019,
     "end_time": "2021-02-03T15:00:11.645511",
     "exception": false,
     "start_time": "2021-02-03T15:00:11.125492",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy: 68%\n"
     ]
    }
   ],
   "source": [
    "model_acc = model.evaluate(X_test, y_test, verbose=0)[1]\n",
    "\n",
    "print(\"Test Accuracy: {:.0f}%\".format(model_acc * 100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T15:00:11.911300Z",
     "iopub.status.busy": "2021-02-03T15:00:11.910531Z",
     "iopub.status.idle": "2021-02-03T15:00:12.366123Z",
     "shell.execute_reply": "2021-02-03T15:00:12.366676Z"
    },
    "papermill": {
     "duration": 0.591938,
     "end_time": "2021-02-03T15:00:12.366874",
     "exception": false,
     "start_time": "2021-02-03T15:00:11.774936",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "y_true = np.array(y_test)\n",
    "\n",
    "y_pred = model.predict(X_test)\n",
    "y_pred = np.array(list(map(lambda x: np.argmax(x), y_pred)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T15:00:12.624522Z",
     "iopub.status.busy": "2021-02-03T15:00:12.623781Z",
     "iopub.status.idle": "2021-02-03T15:00:12.649358Z",
     "shell.execute_reply": "2021-02-03T15:00:12.648792Z"
    },
    "papermill": {
     "duration": 0.155808,
     "end_time": "2021-02-03T15:00:12.649512",
     "exception": false,
     "start_time": "2021-02-03T15:00:12.493704",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.7/site-packages/sklearn/metrics/_classification.py:1221: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n",
      "  _warn_prf(average, modifier, msg_start, len(result))\n"
     ]
    }
   ],
   "source": [
    "cm = confusion_matrix(y_true, y_pred)\n",
    "clr = classification_report(y_true, y_pred, target_names=label_mapping.values())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T15:00:12.934244Z",
     "iopub.status.busy": "2021-02-03T15:00:12.916161Z",
     "iopub.status.idle": "2021-02-03T15:00:13.321482Z",
     "shell.execute_reply": "2021-02-03T15:00:13.322012Z"
    },
    "papermill": {
     "duration": 0.544703,
     "end_time": "2021-02-03T15:00:13.322191",
     "exception": false,
     "start_time": "2021-02-03T15:00:12.777488",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 10))\n",
    "sns.heatmap(cm, annot=True, fmt='g', vmin=0, cbar=False, cmap='Blues')\n",
    "\n",
    "plt.xticks(np.arange(7) + 0.5, label_mapping.values())\n",
    "plt.xlabel(\"Predicted\")\n",
    "\n",
    "plt.yticks(np.arange(7) + 0.5, label_mapping.values())\n",
    "plt.ylabel(\"Actual\")\n",
    "\n",
    "plt.title(\"Confusion Matrix\")\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T15:00:13.583418Z",
     "iopub.status.busy": "2021-02-03T15:00:13.582717Z",
     "iopub.status.idle": "2021-02-03T15:00:13.586508Z",
     "shell.execute_reply": "2021-02-03T15:00:13.585627Z"
    },
    "papermill": {
     "duration": 0.136197,
     "end_time": "2021-02-03T15:00:13.586701",
     "exception": false,
     "start_time": "2021-02-03T15:00:13.450504",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "          nv       0.20      0.01      0.02        94\n",
      "         mel       0.42      0.10      0.17       154\n",
      "         bkl       0.38      0.09      0.14       328\n",
      "         bcc       0.00      0.00      0.00        37\n",
      "       akiec       0.70      0.97      0.81      1990\n",
      "        vasc       0.00      0.00      0.00        49\n",
      "          df       0.41      0.13      0.20       353\n",
      "\n",
      "    accuracy                           0.68      3005\n",
      "   macro avg       0.30      0.19      0.19      3005\n",
      "weighted avg       0.58      0.68      0.59      3005\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(clr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "execution": {
     "iopub.execute_input": "2021-02-03T15:00:13.852750Z",
     "iopub.status.busy": "2021-02-03T15:00:13.850798Z",
     "iopub.status.idle": "2021-02-03T15:00:15.171751Z",
     "shell.execute_reply": "2021-02-03T15:00:15.170958Z"
    },
    "papermill": {
     "duration": 1.455185,
     "end_time": "2021-02-03T15:00:15.171900",
     "exception": false,
     "start_time": "2021-02-03T15:00:13.716715",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "converter = tf.lite.TFLiteConverter.from_keras_model(model)\n",
    "tflite_model = converter.convert()\n",
    "\n",
    "# Save the model.\n",
    "with open('model.tflite', 'wb') as f:\n",
    "  f.write(tflite_model)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 45.077869,
   "end_time": "2021-02-03T15:00:16.512959",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2021-02-03T14:59:31.435090",
   "version": "2.2.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}