Epileptic_seizure_recognition.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "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": "3.6.8"
    },
    "colab": {
      "name": "Epileptic_seizure_recognition.ipynb",
      "provenance": [],
      "collapsed_sections": []
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pivvD9iVDgIW",
        "colab_type": "text"
      },
      "source": [
        "# Epileptic Seizure Recognition with Wavelet Transform\n",
        "\n",
        "#### Nur Altıparmak\n",
        "Epileptic seizure is a disease that affects a large number of people and it is caused by abnormal electrical activity in the brain [1]. Most seizures are very brief and rarely life threatening but critical seizures have to be recognized immediately.Also, according to studies[6], it is possible to control the attacks with early diagnosis.In addition Epileptic seizure patients require special treatments and continuous observation for the disease symptoms. Therefore several methods for epileptic seizure recognition have been developed in the past for early diagnosis. One of them is recognition with EEG signals. Even though EEG signals are a strong tool for epileptic seizure recognition, visual inspect for discriminating EEGs is a time consuming and high costly process [2].Therefore this project is aimed to make epileptic seizure detection using EEG signals that are transformed by wavelet transform, with machine learning and deep learning techniques.  "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0G8qnTHGDgIg",
        "colab_type": "text"
      },
      "source": [
        "## Table of Content\n",
        "\n",
        "[Problem](#problem)   \n",
        "[Data Understanding](#data_understanding)   \n",
        "[Data Preparation](#data_preparation)   \n",
        "[Modeling](#modeling)   \n",
        "[Evaluation](#evaluation)   \n",
        "[References](#references)   \n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GDI1hBUSDgIt",
        "colab_type": "text"
      },
      "source": [
        "## Problem <a class=\"anchor\" id=\"problem\"></a>\n",
        "\n",
        "Epilepsy is a group of neurological disorders characterized by recurrent epileptic seizures.[3]This seizures can cause severe shaking attacks therefore in these seizures patients can hurt her/hisself very badly.Luckily epilepsy is a curable diasease that 70% of cases are controlable with early diagnosis and medication[4][5].Therefore in order to make an early diagnosis, many epilepsy diagnostic methods have been developed over time. One of them is diagnosis with the help of EEG signals.However the visual inspect process for discriminating EEGs is a time consuming and high costly process.Therefore the project is aimed to make epileptic seizure detection very quickly and accurately by using EEG signal of the brain data with machine learning and deep learning techniques."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dh2aefCnDgJB",
        "colab_type": "text"
      },
      "source": [
        "## Data Understanding<a class=\"anchor\" id=\"data_understanding\"></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "aSE-68RTfkDr",
        "colab_type": "text"
      },
      "source": [
        "For this task,I will use Bonn University’s restructured EEG time series dataset that consists of 5 different target classes , 179 attributes and it contains 11500 samples.The classes 2, 3, 4, and 5 are subjects who did not have epileptic seizure. Only subjects in class 1 have epileptic seizure.I provied more details about dataset below.\n",
        "\n",
        "![alt text](https://drive.google.com/uc?id=1nkOyWgqc1jc7LXD-aOfzOoA_cQa7eJ0p)\n",
        "\n",
        "$$$$\n",
        "\n",
        "![alt text](https://drive.google.com/uc?id=1UrYrwvOzx0w57axwOJL7FvJ16M-d2V8B)\n",
        "\n",
        "This dataset is available on https://archive.ics.uci.edu/ml/datasets/Epileptic+Seizure+Recognition"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k7qU8egaDgJB",
        "colab_type": "code",
        "outputId": "7306bb1a-5672-4407-8e5f-7808f3b565bf",
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 74
        }
      },
      "source": [
        "from google.colab import files\n",
        "uploaded = files.upload() #choose data.csv file from provided files"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-10b9dc7b-a609-4b16-afe0-7b5f8cf7bfe1\" name=\"files[]\" multiple disabled />\n",
              "     <output id=\"result-10b9dc7b-a609-4b16-afe0-7b5f8cf7bfe1\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving data.csv to data.csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FGBgpHlADgJF",
        "colab_type": "code",
        "outputId": "0864c506-6ba8-45f5-d484-1d467013af46",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 261
        }
      },
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "df=pd.read_csv(\"data.csv\")\n",
        "df.head()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Unnamed: 0</th>\n",
              "      <th>X1</th>\n",
              "      <th>X2</th>\n",
              "      <th>X3</th>\n",
              "      <th>X4</th>\n",
              "      <th>X5</th>\n",
              "      <th>X6</th>\n",
              "      <th>X7</th>\n",
              "      <th>X8</th>\n",
              "      <th>X9</th>\n",
              "      <th>X10</th>\n",
              "      <th>X11</th>\n",
              "      <th>X12</th>\n",
              "      <th>X13</th>\n",
              "      <th>X14</th>\n",
              "      <th>X15</th>\n",
              "      <th>X16</th>\n",
              "      <th>X17</th>\n",
              "      <th>X18</th>\n",
              "      <th>X19</th>\n",
              "      <th>X20</th>\n",
              "      <th>X21</th>\n",
              "      <th>X22</th>\n",
              "      <th>X23</th>\n",
              "      <th>X24</th>\n",
              "      <th>X25</th>\n",
              "      <th>X26</th>\n",
              "      <th>X27</th>\n",
              "      <th>X28</th>\n",
              "      <th>X29</th>\n",
              "      <th>X30</th>\n",
              "      <th>X31</th>\n",
              "      <th>X32</th>\n",
              "      <th>X33</th>\n",
              "      <th>X34</th>\n",
              "      <th>X35</th>\n",
              "      <th>X36</th>\n",
              "      <th>X37</th>\n",
              "      <th>X38</th>\n",
              "      <th>X39</th>\n",
              "      <th>...</th>\n",
              "      <th>X140</th>\n",
              "      <th>X141</th>\n",
              "      <th>X142</th>\n",
              "      <th>X143</th>\n",
              "      <th>X144</th>\n",
              "      <th>X145</th>\n",
              "      <th>X146</th>\n",
              "      <th>X147</th>\n",
              "      <th>X148</th>\n",
              "      <th>X149</th>\n",
              "      <th>X150</th>\n",
              "      <th>X151</th>\n",
              "      <th>X152</th>\n",
              "      <th>X153</th>\n",
              "      <th>X154</th>\n",
              "      <th>X155</th>\n",
              "      <th>X156</th>\n",
              "      <th>X157</th>\n",
              "      <th>X158</th>\n",
              "      <th>X159</th>\n",
              "      <th>X160</th>\n",
              "      <th>X161</th>\n",
              "      <th>X162</th>\n",
              "      <th>X163</th>\n",
              "      <th>X164</th>\n",
              "      <th>X165</th>\n",
              "      <th>X166</th>\n",
              "      <th>X167</th>\n",
              "      <th>X168</th>\n",
              "      <th>X169</th>\n",
              "      <th>X170</th>\n",
              "      <th>X171</th>\n",
              "      <th>X172</th>\n",
              "      <th>X173</th>\n",
              "      <th>X174</th>\n",
              "      <th>X175</th>\n",
              "      <th>X176</th>\n",
              "      <th>X177</th>\n",
              "      <th>X178</th>\n",
              "      <th>y</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>X21.V1.791</td>\n",
              "      <td>135</td>\n",
              "      <td>190</td>\n",
              "      <td>229</td>\n",
              "      <td>223</td>\n",
              "      <td>192</td>\n",
              "      <td>125</td>\n",
              "      <td>55</td>\n",
              "      <td>-9</td>\n",
              "      <td>-33</td>\n",
              "      <td>-38</td>\n",
              "      <td>-10</td>\n",
              "      <td>35</td>\n",
              "      <td>64</td>\n",
              "      <td>113</td>\n",
              "      <td>152</td>\n",
              "      <td>164</td>\n",
              "      <td>127</td>\n",
              "      <td>50</td>\n",
              "      <td>-47</td>\n",
              "      <td>-121</td>\n",
              "      <td>-138</td>\n",
              "      <td>-125</td>\n",
              "      <td>-101</td>\n",
              "      <td>-50</td>\n",
              "      <td>11</td>\n",
              "      <td>39</td>\n",
              "      <td>24</td>\n",
              "      <td>48</td>\n",
              "      <td>64</td>\n",
              "      <td>46</td>\n",
              "      <td>13</td>\n",
              "      <td>-19</td>\n",
              "      <td>-61</td>\n",
              "      <td>-96</td>\n",
              "      <td>-130</td>\n",
              "      <td>-132</td>\n",
              "      <td>-116</td>\n",
              "      <td>-115</td>\n",
              "      <td>-71</td>\n",
              "      <td>...</td>\n",
              "      <td>54</td>\n",
              "      <td>90</td>\n",
              "      <td>111</td>\n",
              "      <td>107</td>\n",
              "      <td>64</td>\n",
              "      <td>32</td>\n",
              "      <td>18</td>\n",
              "      <td>-25</td>\n",
              "      <td>-69</td>\n",
              "      <td>-65</td>\n",
              "      <td>-44</td>\n",
              "      <td>-33</td>\n",
              "      <td>-57</td>\n",
              "      <td>-88</td>\n",
              "      <td>-114</td>\n",
              "      <td>-130</td>\n",
              "      <td>-114</td>\n",
              "      <td>-83</td>\n",
              "      <td>-53</td>\n",
              "      <td>-79</td>\n",
              "      <td>-72</td>\n",
              "      <td>-85</td>\n",
              "      <td>-109</td>\n",
              "      <td>-98</td>\n",
              "      <td>-72</td>\n",
              "      <td>-65</td>\n",
              "      <td>-63</td>\n",
              "      <td>-11</td>\n",
              "      <td>10</td>\n",
              "      <td>8</td>\n",
              "      <td>-17</td>\n",
              "      <td>-15</td>\n",
              "      <td>-31</td>\n",
              "      <td>-77</td>\n",
              "      <td>-103</td>\n",
              "      <td>-127</td>\n",
              "      <td>-116</td>\n",
              "      <td>-83</td>\n",
              "      <td>-51</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>X15.V1.924</td>\n",
              "      <td>386</td>\n",
              "      <td>382</td>\n",
              "      <td>356</td>\n",
              "      <td>331</td>\n",
              "      <td>320</td>\n",
              "      <td>315</td>\n",
              "      <td>307</td>\n",
              "      <td>272</td>\n",
              "      <td>244</td>\n",
              "      <td>232</td>\n",
              "      <td>237</td>\n",
              "      <td>258</td>\n",
              "      <td>212</td>\n",
              "      <td>2</td>\n",
              "      <td>-267</td>\n",
              "      <td>-605</td>\n",
              "      <td>-850</td>\n",
              "      <td>-1001</td>\n",
              "      <td>-1109</td>\n",
              "      <td>-1090</td>\n",
              "      <td>-967</td>\n",
              "      <td>-746</td>\n",
              "      <td>-464</td>\n",
              "      <td>-152</td>\n",
              "      <td>118</td>\n",
              "      <td>318</td>\n",
              "      <td>427</td>\n",
              "      <td>473</td>\n",
              "      <td>485</td>\n",
              "      <td>447</td>\n",
              "      <td>397</td>\n",
              "      <td>339</td>\n",
              "      <td>312</td>\n",
              "      <td>314</td>\n",
              "      <td>326</td>\n",
              "      <td>335</td>\n",
              "      <td>332</td>\n",
              "      <td>324</td>\n",
              "      <td>310</td>\n",
              "      <td>...</td>\n",
              "      <td>27</td>\n",
              "      <td>146</td>\n",
              "      <td>229</td>\n",
              "      <td>269</td>\n",
              "      <td>297</td>\n",
              "      <td>307</td>\n",
              "      <td>303</td>\n",
              "      <td>305</td>\n",
              "      <td>306</td>\n",
              "      <td>307</td>\n",
              "      <td>280</td>\n",
              "      <td>231</td>\n",
              "      <td>159</td>\n",
              "      <td>85</td>\n",
              "      <td>51</td>\n",
              "      <td>43</td>\n",
              "      <td>62</td>\n",
              "      <td>63</td>\n",
              "      <td>63</td>\n",
              "      <td>69</td>\n",
              "      <td>89</td>\n",
              "      <td>123</td>\n",
              "      <td>136</td>\n",
              "      <td>127</td>\n",
              "      <td>102</td>\n",
              "      <td>95</td>\n",
              "      <td>105</td>\n",
              "      <td>131</td>\n",
              "      <td>163</td>\n",
              "      <td>168</td>\n",
              "      <td>164</td>\n",
              "      <td>150</td>\n",
              "      <td>146</td>\n",
              "      <td>152</td>\n",
              "      <td>157</td>\n",
              "      <td>156</td>\n",
              "      <td>154</td>\n",
              "      <td>143</td>\n",
              "      <td>129</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>X8.V1.1</td>\n",
              "      <td>-32</td>\n",
              "      <td>-39</td>\n",
              "      <td>-47</td>\n",
              "      <td>-37</td>\n",
              "      <td>-32</td>\n",
              "      <td>-36</td>\n",
              "      <td>-57</td>\n",
              "      <td>-73</td>\n",
              "      <td>-85</td>\n",
              "      <td>-94</td>\n",
              "      <td>-99</td>\n",
              "      <td>-94</td>\n",
              "      <td>-96</td>\n",
              "      <td>-104</td>\n",
              "      <td>-103</td>\n",
              "      <td>-92</td>\n",
              "      <td>-75</td>\n",
              "      <td>-69</td>\n",
              "      <td>-69</td>\n",
              "      <td>-53</td>\n",
              "      <td>-37</td>\n",
              "      <td>-14</td>\n",
              "      <td>-10</td>\n",
              "      <td>-39</td>\n",
              "      <td>-78</td>\n",
              "      <td>-102</td>\n",
              "      <td>-98</td>\n",
              "      <td>-80</td>\n",
              "      <td>-54</td>\n",
              "      <td>-40</td>\n",
              "      <td>-35</td>\n",
              "      <td>-39</td>\n",
              "      <td>-32</td>\n",
              "      <td>-13</td>\n",
              "      <td>7</td>\n",
              "      <td>34</td>\n",
              "      <td>41</td>\n",
              "      <td>33</td>\n",
              "      <td>6</td>\n",
              "      <td>...</td>\n",
              "      <td>-82</td>\n",
              "      <td>-107</td>\n",
              "      <td>-126</td>\n",
              "      <td>-124</td>\n",
              "      <td>-108</td>\n",
              "      <td>-84</td>\n",
              "      <td>-68</td>\n",
              "      <td>-61</td>\n",
              "      <td>-56</td>\n",
              "      <td>-63</td>\n",
              "      <td>-62</td>\n",
              "      <td>-33</td>\n",
              "      <td>1</td>\n",
              "      <td>28</td>\n",
              "      <td>45</td>\n",
              "      <td>37</td>\n",
              "      <td>48</td>\n",
              "      <td>62</td>\n",
              "      <td>80</td>\n",
              "      <td>66</td>\n",
              "      <td>23</td>\n",
              "      <td>-11</td>\n",
              "      <td>-39</td>\n",
              "      <td>-44</td>\n",
              "      <td>-42</td>\n",
              "      <td>-45</td>\n",
              "      <td>-48</td>\n",
              "      <td>-42</td>\n",
              "      <td>-6</td>\n",
              "      <td>29</td>\n",
              "      <td>57</td>\n",
              "      <td>64</td>\n",
              "      <td>48</td>\n",
              "      <td>19</td>\n",
              "      <td>-12</td>\n",
              "      <td>-30</td>\n",
              "      <td>-35</td>\n",
              "      <td>-35</td>\n",
              "      <td>-36</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>X16.V1.60</td>\n",
              "      <td>-105</td>\n",
              "      <td>-101</td>\n",
              "      <td>-96</td>\n",
              "      <td>-92</td>\n",
              "      <td>-89</td>\n",
              "      <td>-95</td>\n",
              "      <td>-102</td>\n",
              "      <td>-100</td>\n",
              "      <td>-87</td>\n",
              "      <td>-79</td>\n",
              "      <td>-72</td>\n",
              "      <td>-68</td>\n",
              "      <td>-74</td>\n",
              "      <td>-80</td>\n",
              "      <td>-83</td>\n",
              "      <td>-73</td>\n",
              "      <td>-68</td>\n",
              "      <td>-61</td>\n",
              "      <td>-58</td>\n",
              "      <td>-59</td>\n",
              "      <td>-64</td>\n",
              "      <td>-79</td>\n",
              "      <td>-84</td>\n",
              "      <td>-97</td>\n",
              "      <td>-94</td>\n",
              "      <td>-84</td>\n",
              "      <td>-77</td>\n",
              "      <td>-75</td>\n",
              "      <td>-72</td>\n",
              "      <td>-68</td>\n",
              "      <td>-76</td>\n",
              "      <td>-76</td>\n",
              "      <td>-72</td>\n",
              "      <td>-67</td>\n",
              "      <td>-69</td>\n",
              "      <td>-69</td>\n",
              "      <td>-69</td>\n",
              "      <td>-67</td>\n",
              "      <td>-68</td>\n",
              "      <td>...</td>\n",
              "      <td>-69</td>\n",
              "      <td>-66</td>\n",
              "      <td>-74</td>\n",
              "      <td>-69</td>\n",
              "      <td>-61</td>\n",
              "      <td>-51</td>\n",
              "      <td>-45</td>\n",
              "      <td>-45</td>\n",
              "      <td>-49</td>\n",
              "      <td>-58</td>\n",
              "      <td>-64</td>\n",
              "      <td>-78</td>\n",
              "      <td>-80</td>\n",
              "      <td>-90</td>\n",
              "      <td>-87</td>\n",
              "      <td>-83</td>\n",
              "      <td>-78</td>\n",
              "      <td>-64</td>\n",
              "      <td>-38</td>\n",
              "      <td>-22</td>\n",
              "      <td>-29</td>\n",
              "      <td>-42</td>\n",
              "      <td>-51</td>\n",
              "      <td>-68</td>\n",
              "      <td>-71</td>\n",
              "      <td>-69</td>\n",
              "      <td>-69</td>\n",
              "      <td>-74</td>\n",
              "      <td>-74</td>\n",
              "      <td>-80</td>\n",
              "      <td>-82</td>\n",
              "      <td>-81</td>\n",
              "      <td>-80</td>\n",
              "      <td>-77</td>\n",
              "      <td>-85</td>\n",
              "      <td>-77</td>\n",
              "      <td>-72</td>\n",
              "      <td>-69</td>\n",
              "      <td>-65</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>X20.V1.54</td>\n",
              "      <td>-9</td>\n",
              "      <td>-65</td>\n",
              "      <td>-98</td>\n",
              "      <td>-102</td>\n",
              "      <td>-78</td>\n",
              "      <td>-48</td>\n",
              "      <td>-16</td>\n",
              "      <td>0</td>\n",
              "      <td>-21</td>\n",
              "      <td>-59</td>\n",
              "      <td>-90</td>\n",
              "      <td>-103</td>\n",
              "      <td>-84</td>\n",
              "      <td>-43</td>\n",
              "      <td>-9</td>\n",
              "      <td>3</td>\n",
              "      <td>-21</td>\n",
              "      <td>-60</td>\n",
              "      <td>-96</td>\n",
              "      <td>-103</td>\n",
              "      <td>-75</td>\n",
              "      <td>-29</td>\n",
              "      <td>14</td>\n",
              "      <td>55</td>\n",
              "      <td>78</td>\n",
              "      <td>73</td>\n",
              "      <td>28</td>\n",
              "      <td>-13</td>\n",
              "      <td>-43</td>\n",
              "      <td>-68</td>\n",
              "      <td>-78</td>\n",
              "      <td>-75</td>\n",
              "      <td>-55</td>\n",
              "      <td>-41</td>\n",
              "      <td>-19</td>\n",
              "      <td>-20</td>\n",
              "      <td>-29</td>\n",
              "      <td>-36</td>\n",
              "      <td>-20</td>\n",
              "      <td>...</td>\n",
              "      <td>38</td>\n",
              "      <td>26</td>\n",
              "      <td>10</td>\n",
              "      <td>-4</td>\n",
              "      <td>-13</td>\n",
              "      <td>-8</td>\n",
              "      <td>0</td>\n",
              "      <td>10</td>\n",
              "      <td>19</td>\n",
              "      <td>29</td>\n",
              "      <td>57</td>\n",
              "      <td>63</td>\n",
              "      <td>45</td>\n",
              "      <td>7</td>\n",
              "      <td>-13</td>\n",
              "      <td>-23</td>\n",
              "      <td>-9</td>\n",
              "      <td>9</td>\n",
              "      <td>11</td>\n",
              "      <td>3</td>\n",
              "      <td>-1</td>\n",
              "      <td>-2</td>\n",
              "      <td>4</td>\n",
              "      <td>18</td>\n",
              "      <td>27</td>\n",
              "      <td>27</td>\n",
              "      <td>14</td>\n",
              "      <td>15</td>\n",
              "      <td>11</td>\n",
              "      <td>10</td>\n",
              "      <td>4</td>\n",
              "      <td>2</td>\n",
              "      <td>-12</td>\n",
              "      <td>-32</td>\n",
              "      <td>-41</td>\n",
              "      <td>-65</td>\n",
              "      <td>-83</td>\n",
              "      <td>-89</td>\n",
              "      <td>-73</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>5 rows × 180 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "   Unnamed: 0   X1   X2   X3   X4   X5  ...  X174  X175  X176  X177  X178  y\n",
              "0  X21.V1.791  135  190  229  223  192  ...  -103  -127  -116   -83   -51  4\n",
              "1  X15.V1.924  386  382  356  331  320  ...   157   156   154   143   129  1\n",
              "2     X8.V1.1  -32  -39  -47  -37  -32  ...   -12   -30   -35   -35   -36  5\n",
              "3   X16.V1.60 -105 -101  -96  -92  -89  ...   -85   -77   -72   -69   -65  5\n",
              "4   X20.V1.54   -9  -65  -98 -102  -78  ...   -41   -65   -83   -89   -73  5\n",
              "\n",
              "[5 rows x 180 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cHV2pSs3ZYsL",
        "colab_type": "code",
        "outputId": "b9f23367-1ab6-481d-b1a5-42f4626a0d26",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 252
        }
      },
      "source": [
        "print(\"General info about colums,rows etc.\")\n",
        "df.info()\n",
        "print(\"\\nTarget variables value counts\\n\",df[\"y\"].value_counts())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "General info about colums,rows etc.\n",
            "<class 'pandas.core.frame.DataFrame'>\n",
            "RangeIndex: 11500 entries, 0 to 11499\n",
            "Columns: 180 entries, Unnamed: 0 to y\n",
            "dtypes: int64(179), object(1)\n",
            "memory usage: 15.8+ MB\n",
            "\n",
            "Target variables value counts\n",
            " 5    2300\n",
            "4    2300\n",
            "3    2300\n",
            "2    2300\n",
            "1    2300\n",
            "Name: y, dtype: int64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "03cZ591QaUWD",
        "colab_type": "code",
        "outputId": "1eae8843-30fa-450f-e526-30ec82b2c04f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "def hist(df,plt):\n",
        "  plt.hist(df[df[\"y\"]==1][\"y\"],label=\"epileptic seizure activity\")\n",
        "  plt.hist(df[df[\"y\"]!=1][\"y\"],label=\"not a seizure\")\n",
        "  plt.legend(loc='lower right')\n",
        "  plt.show()\n",
        "\n",
        "hist(df,plt)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ehj245a_a65A",
        "colab_type": "text"
      },
      "source": [
        "All class have same amount of instances and dataset looks balance however i will transform target variable into seizure or not so dataset will no longer be balanced."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hIg2VoBTDgJL",
        "colab_type": "text"
      },
      "source": [
        "## Data Preparation<a class=\"anchor\" id=\"data_preparation\"></a>\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4dnmF33egWOV",
        "colab_type": "text"
      },
      "source": [
        "In data preparation process, firstly i cleaned the data from unnecessary columns then I transformed the target variable into binary classes (1 for seizure, 0 for not seizure).Then I created new variables with some mathematical transformations that called Hurst exponent and Discrete wavelet transformation (do not worry, I will explain all these transformations below.). Then i normalize data.After i restructured the unbalanced data into more balanced data.Lastly I inspected whether feature selection improve accuracies."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BU53DhA_mshB",
        "colab_type": "text"
      },
      "source": [
        "######**Cleaning the data and transforming the target variable**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XPUtD2EtogJ2",
        "colab_type": "code",
        "outputId": "57eaaf22-3226-4ee0-c7a5-beb3dc8d618e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 172
        }
      },
      "source": [
        "df.head(2) #just a quick look to data again. x1...x178 columns are a part of our time-series data (EEG Signals) but Unnamed: 0 column must be inspected."
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>Unnamed: 0</th>\n",
              "      <th>X1</th>\n",
              "      <th>X2</th>\n",
              "      <th>X3</th>\n",
              "      <th>X4</th>\n",
              "      <th>X5</th>\n",
              "      <th>X6</th>\n",
              "      <th>X7</th>\n",
              "      <th>X8</th>\n",
              "      <th>X9</th>\n",
              "      <th>X10</th>\n",
              "      <th>X11</th>\n",
              "      <th>X12</th>\n",
              "      <th>X13</th>\n",
              "      <th>X14</th>\n",
              "      <th>X15</th>\n",
              "      <th>X16</th>\n",
              "      <th>X17</th>\n",
              "      <th>X18</th>\n",
              "      <th>X19</th>\n",
              "      <th>X20</th>\n",
              "      <th>X21</th>\n",
              "      <th>X22</th>\n",
              "      <th>X23</th>\n",
              "      <th>X24</th>\n",
              "      <th>X25</th>\n",
              "      <th>X26</th>\n",
              "      <th>X27</th>\n",
              "      <th>X28</th>\n",
              "      <th>X29</th>\n",
              "      <th>X30</th>\n",
              "      <th>X31</th>\n",
              "      <th>X32</th>\n",
              "      <th>X33</th>\n",
              "      <th>X34</th>\n",
              "      <th>X35</th>\n",
              "      <th>X36</th>\n",
              "      <th>X37</th>\n",
              "      <th>X38</th>\n",
              "      <th>X39</th>\n",
              "      <th>...</th>\n",
              "      <th>X140</th>\n",
              "      <th>X141</th>\n",
              "      <th>X142</th>\n",
              "      <th>X143</th>\n",
              "      <th>X144</th>\n",
              "      <th>X145</th>\n",
              "      <th>X146</th>\n",
              "      <th>X147</th>\n",
              "      <th>X148</th>\n",
              "      <th>X149</th>\n",
              "      <th>X150</th>\n",
              "      <th>X151</th>\n",
              "      <th>X152</th>\n",
              "      <th>X153</th>\n",
              "      <th>X154</th>\n",
              "      <th>X155</th>\n",
              "      <th>X156</th>\n",
              "      <th>X157</th>\n",
              "      <th>X158</th>\n",
              "      <th>X159</th>\n",
              "      <th>X160</th>\n",
              "      <th>X161</th>\n",
              "      <th>X162</th>\n",
              "      <th>X163</th>\n",
              "      <th>X164</th>\n",
              "      <th>X165</th>\n",
              "      <th>X166</th>\n",
              "      <th>X167</th>\n",
              "      <th>X168</th>\n",
              "      <th>X169</th>\n",
              "      <th>X170</th>\n",
              "      <th>X171</th>\n",
              "      <th>X172</th>\n",
              "      <th>X173</th>\n",
              "      <th>X174</th>\n",
              "      <th>X175</th>\n",
              "      <th>X176</th>\n",
              "      <th>X177</th>\n",
              "      <th>X178</th>\n",
              "      <th>y</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>X21.V1.791</td>\n",
              "      <td>135</td>\n",
              "      <td>190</td>\n",
              "      <td>229</td>\n",
              "      <td>223</td>\n",
              "      <td>192</td>\n",
              "      <td>125</td>\n",
              "      <td>55</td>\n",
              "      <td>-9</td>\n",
              "      <td>-33</td>\n",
              "      <td>-38</td>\n",
              "      <td>-10</td>\n",
              "      <td>35</td>\n",
              "      <td>64</td>\n",
              "      <td>113</td>\n",
              "      <td>152</td>\n",
              "      <td>164</td>\n",
              "      <td>127</td>\n",
              "      <td>50</td>\n",
              "      <td>-47</td>\n",
              "      <td>-121</td>\n",
              "      <td>-138</td>\n",
              "      <td>-125</td>\n",
              "      <td>-101</td>\n",
              "      <td>-50</td>\n",
              "      <td>11</td>\n",
              "      <td>39</td>\n",
              "      <td>24</td>\n",
              "      <td>48</td>\n",
              "      <td>64</td>\n",
              "      <td>46</td>\n",
              "      <td>13</td>\n",
              "      <td>-19</td>\n",
              "      <td>-61</td>\n",
              "      <td>-96</td>\n",
              "      <td>-130</td>\n",
              "      <td>-132</td>\n",
              "      <td>-116</td>\n",
              "      <td>-115</td>\n",
              "      <td>-71</td>\n",
              "      <td>...</td>\n",
              "      <td>54</td>\n",
              "      <td>90</td>\n",
              "      <td>111</td>\n",
              "      <td>107</td>\n",
              "      <td>64</td>\n",
              "      <td>32</td>\n",
              "      <td>18</td>\n",
              "      <td>-25</td>\n",
              "      <td>-69</td>\n",
              "      <td>-65</td>\n",
              "      <td>-44</td>\n",
              "      <td>-33</td>\n",
              "      <td>-57</td>\n",
              "      <td>-88</td>\n",
              "      <td>-114</td>\n",
              "      <td>-130</td>\n",
              "      <td>-114</td>\n",
              "      <td>-83</td>\n",
              "      <td>-53</td>\n",
              "      <td>-79</td>\n",
              "      <td>-72</td>\n",
              "      <td>-85</td>\n",
              "      <td>-109</td>\n",
              "      <td>-98</td>\n",
              "      <td>-72</td>\n",
              "      <td>-65</td>\n",
              "      <td>-63</td>\n",
              "      <td>-11</td>\n",
              "      <td>10</td>\n",
              "      <td>8</td>\n",
              "      <td>-17</td>\n",
              "      <td>-15</td>\n",
              "      <td>-31</td>\n",
              "      <td>-77</td>\n",
              "      <td>-103</td>\n",
              "      <td>-127</td>\n",
              "      <td>-116</td>\n",
              "      <td>-83</td>\n",
              "      <td>-51</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>X15.V1.924</td>\n",
              "      <td>386</td>\n",
              "      <td>382</td>\n",
              "      <td>356</td>\n",
              "      <td>331</td>\n",
              "      <td>320</td>\n",
              "      <td>315</td>\n",
              "      <td>307</td>\n",
              "      <td>272</td>\n",
              "      <td>244</td>\n",
              "      <td>232</td>\n",
              "      <td>237</td>\n",
              "      <td>258</td>\n",
              "      <td>212</td>\n",
              "      <td>2</td>\n",
              "      <td>-267</td>\n",
              "      <td>-605</td>\n",
              "      <td>-850</td>\n",
              "      <td>-1001</td>\n",
              "      <td>-1109</td>\n",
              "      <td>-1090</td>\n",
              "      <td>-967</td>\n",
              "      <td>-746</td>\n",
              "      <td>-464</td>\n",
              "      <td>-152</td>\n",
              "      <td>118</td>\n",
              "      <td>318</td>\n",
              "      <td>427</td>\n",
              "      <td>473</td>\n",
              "      <td>485</td>\n",
              "      <td>447</td>\n",
              "      <td>397</td>\n",
              "      <td>339</td>\n",
              "      <td>312</td>\n",
              "      <td>314</td>\n",
              "      <td>326</td>\n",
              "      <td>335</td>\n",
              "      <td>332</td>\n",
              "      <td>324</td>\n",
              "      <td>310</td>\n",
              "      <td>...</td>\n",
              "      <td>27</td>\n",
              "      <td>146</td>\n",
              "      <td>229</td>\n",
              "      <td>269</td>\n",
              "      <td>297</td>\n",
              "      <td>307</td>\n",
              "      <td>303</td>\n",
              "      <td>305</td>\n",
              "      <td>306</td>\n",
              "      <td>307</td>\n",
              "      <td>280</td>\n",
              "      <td>231</td>\n",
              "      <td>159</td>\n",
              "      <td>85</td>\n",
              "      <td>51</td>\n",
              "      <td>43</td>\n",
              "      <td>62</td>\n",
              "      <td>63</td>\n",
              "      <td>63</td>\n",
              "      <td>69</td>\n",
              "      <td>89</td>\n",
              "      <td>123</td>\n",
              "      <td>136</td>\n",
              "      <td>127</td>\n",
              "      <td>102</td>\n",
              "      <td>95</td>\n",
              "      <td>105</td>\n",
              "      <td>131</td>\n",
              "      <td>163</td>\n",
              "      <td>168</td>\n",
              "      <td>164</td>\n",
              "      <td>150</td>\n",
              "      <td>146</td>\n",
              "      <td>152</td>\n",
              "      <td>157</td>\n",
              "      <td>156</td>\n",
              "      <td>154</td>\n",
              "      <td>143</td>\n",
              "      <td>129</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>2 rows × 180 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "   Unnamed: 0   X1   X2   X3   X4   X5  ...  X174  X175  X176  X177  X178  y\n",
              "0  X21.V1.791  135  190  229  223  192  ...  -103  -127  -116   -83   -51  4\n",
              "1  X15.V1.924  386  382  356  331  320  ...   157   156   154   143   129  1\n",
              "\n",
              "[2 rows x 180 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-o6OUhpAnhew",
        "colab_type": "code",
        "outputId": "b3260d17-b4c9-4a03-ea74-8dcae363604d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 218
        }
      },
      "source": [
        "df[\"Unnamed: 0\"].value_counts #As you can see this column is exlusive for all instance\n",
        "                              #So it means, the column has no effect on classification, it is unnecessary"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<bound method IndexOpsMixin.value_counts of 0        X21.V1.791\n",
              "1        X15.V1.924\n",
              "2           X8.V1.1\n",
              "3         X16.V1.60\n",
              "4         X20.V1.54\n",
              "            ...    \n",
              "11495    X22.V1.114\n",
              "11496    X19.V1.354\n",
              "11497      X8.V1.28\n",
              "11498    X10.V1.932\n",
              "11499    X16.V1.210\n",
              "Name: Unnamed: 0, Length: 11500, dtype: object>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CDDm7AS9o8js",
        "colab_type": "code",
        "outputId": "3d4504b4-ea88-4c7c-d813-d656dd503bb2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 118
        }
      },
      "source": [
        "df[\"y\"].value_counts() # I will transform 2,3,4,5 classes to 0, 1 class to 1"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "5    2300\n",
              "4    2300\n",
              "3    2300\n",
              "2    2300\n",
              "1    2300\n",
              "Name: y, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5TW87L4pDgJM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#This method drop the unnecessary column (Unnamed: 0) and transform the target variable\n",
        "def prepareData(df):\n",
        "  df[\"y\"]=[1 if df[\"y\"][i]==1 else 0 for i in range(len(df[\"y\"]))]\n",
        "  target=df[\"y\"]\n",
        "  df_copy=df.drop([\"Unnamed: 0\",\"y\"],axis=1)\n",
        "  return df_copy,target\n",
        "\n",
        "df_copy,target=prepareData(df)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yeDhvje4qDUu",
        "colab_type": "text"
      },
      "source": [
        "######**Data transformation with hurst exponent and discrete wavelet transform**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "txIHEeaAqhn0",
        "colab_type": "text"
      },
      "source": [
        "Before i started coding the project, I had some literature review.According to my review, I decided on not using raw time series data but using transformed data.I created new features with these transformations called hurst exponent and discrete wavelet transformation.$$$$\n",
        "\n",
        "**Hurst Exponent :**\n",
        "The Hurst exponent is used as a measure of long-term memory of time series. It relates to the autocorrelations of the time series, and the rate at which these decrease as the lag between pairs of values increases[7].Basically the main reason why I use this variable is that I think there may be certain fluctuation differences between the classes of time series data.$$$$\n",
        "\n",
        "Hurst exponent formula:\n",
        "\n",
        "![alt text](https://wikimedia.org/api/rest_v1/media/math/render/svg/d32a26032869f8c902eb7a3062309dd88de63b5b)$$$$\n",
        "**Discrete Wavelet Transform :**\n",
        "\n",
        "Discrete Wavelet Transform (DWT) is a kind of transform that any wavelet transform for which the wavelets are discretely sampled[8].(Not an open explanation right? What is a wavelet transform anyway?).\n",
        "\n",
        "**Note:** I used db4 type DWT which is very popular in literature, for this task and also provided a table that shows which types are used in literature.(see\n",
        "figure 2)\n",
        "\n",
        "**Wavelet Transform :**\n",
        "\n",
        "Actually if you already familiar with Fourier Transform, Wavelet transform can be explained more concretely.In Fourier Transform, the transform can understand which frequencies (frequency domain) exist in the signal without any time information(time domain) but Wavelet transform can give more information about both time and frequencies domains.Wavelet transform give more information on frequency domain on small frequencies.On the other hand it give more information about time domain in large frequency values[9].\n",
        "\n",
        "\n",
        "![alt text](https://drive.google.com/uc?id=1dUhbFebAHnD7h_Yq5GvyKGXKinrzPFXo)\n",
        "\n",
        "(Picture is modified version of given link picture taken from [10]]\n",
        "\n",
        "(For more detail about wavelet transform, you can also check [9] reference.)\n",
        "\n",
        "\n",
        "$$$$\n",
        "\n",
        "![alt text](https://drive.google.com/uc?id=1Mgf76Fem5ts2Tntx4GL-oYGVxc-HQBFy)\n",
        "\n",
        "(Table 2, This table taken from [13] research paper)\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OyClaEM9O1GJ",
        "colab_type": "code",
        "outputId": "c0742ca5-842f-47a8-fb13-5807f840bddf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 168
        }
      },
      "source": [
        "pip install hurst #installing hurst module to get hurst values"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting hurst\n",
            "  Downloading https://files.pythonhosted.org/packages/02/4f/d3471ce0dca03a21d4c6640da07a6040c9cc800a937233086b6cea6a7dc2/hurst-0.0.5-py3-none-any.whl\n",
            "Requirement already satisfied: pandas>=0.18 in /usr/local/lib/python3.6/dist-packages (from hurst) (1.0.3)\n",
            "Requirement already satisfied: numpy>=1.10 in /usr/local/lib/python3.6/dist-packages (from hurst) (1.18.4)\n",
            "Requirement already satisfied: python-dateutil>=2.6.1 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.18->hurst) (2.8.1)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.18->hurst) (2018.9)\n",
            "Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.6/dist-packages (from python-dateutil>=2.6.1->pandas>=0.18->hurst) (1.12.0)\n",
            "Installing collected packages: hurst\n",
            "Successfully installed hurst-0.0.5\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y3GOdevmDgJT",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pywt #importing pywt for getting wavelet transform features\n",
        "from hurst import compute_Hc"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TwWSGlqTPKww",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def getHurst(df_copy):\n",
        "  df_copy[\"hurst_ex\"]=[compute_Hc(df_copy.iloc[i], kind=\"change\", simplified=True)[0] for i in range(len(df_copy))]\n",
        "  df_copy[\"hurst_c\"]=[compute_Hc(df_copy.iloc[i], kind=\"change\", simplified=True)[1] for i in range(len(df_copy))]\n",
        "  return df_copy\n",
        "\n",
        "\n",
        "def getStatsForHurst(df_copy):\n",
        "  plt.scatter(df_copy[\"hurst_ex\"],target)\n",
        "  print(\"mean value of hurst exponent for class 1:\",np.mean(df_copy.iloc[target[target==1].index][\"hurst_ex\"]))\n",
        "  print(\"mean value of hurst exponent for class 0:\",np.mean(df_copy.iloc[target[target==0].index][\"hurst_ex\"]))\n",
        "  print(\"mean value of hurst constant for class 1:\",np.mean(df_copy.iloc[target[target==1].index][\"hurst_c\"]))\n",
        "  print(\"mean value of hurst constant for class 0:\",np.mean(df_copy.iloc[target[target==0].index][\"hurst_c\"]))\n",
        "  print(\"median value of hurst exponent for class 1:\",np.median(df_copy.iloc[target[target==1].index][\"hurst_ex\"]))\n",
        "  print(\"median value of hurst exponent for class 0:\",np.median(df_copy.iloc[target[target==0].index][\"hurst_ex\"]))\n",
        "  print(\"median value of hurst constant for class 1:\",np.median(df_copy.iloc[target[target==1].index][\"hurst_c\"]))\n",
        "  print(\"median value of hurst constant for class 0:\",np.median(df_copy.iloc[target[target==0].index][\"hurst_c\"]))\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DN6ZyHfePWx5",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#These methods create a new dataset with wavelet transform\n",
        "#In getWaveletFeatures method, i get a group of wavelet coeffient and hurst exponent and the constant for all instance \n",
        "#give these values to statisticsForWavelet function to get coeffients quartiles,mean,median,standart deviation,variance,root mean square and some other values.\n",
        "#Lastly createDfWavelet method give all these values and return a new dataframe\n",
        "def getWaveletFeatures(data,target):\n",
        "    list_features = []\n",
        "    for signal in range(len(data)):\n",
        "        list_coeff = pywt.wavedec(data.iloc[signal], \"db4\")\n",
        "        features = []\n",
        "        features.append(data.iloc[signal][\"hurst_ex\"])\n",
        "        features.append(data.iloc[signal][\"hurst_c\"])\n",
        "        for coeff in list_coeff:\n",
        "            features += statisticsForWavelet(coeff)\n",
        "        list_features.append(features)\n",
        "    return createDfWavelet(list_features,target)\n",
        "#This method taken from [9]\n",
        "def statisticsForWavelet(coefs):\n",
        "    n5 = np.nanpercentile(coefs, 5)\n",
        "    n25 = np.nanpercentile(coefs, 25)\n",
        "    n75 = np.nanpercentile(coefs, 75)\n",
        "    n95 = np.nanpercentile(coefs, 95)\n",
        "    median = np.nanpercentile(coefs, 50)\n",
        "    mean = np.nanmean(coefs)\n",
        "    std = np.nanstd(coefs)\n",
        "    var = np.nanvar(coefs)\n",
        "    rms = np.nanmean(np.sqrt(coefs**2))\n",
        "    return [n5, n25, n75, n95, median, mean, std, var, rms]\n",
        "\n",
        "def createDfWavelet(data,target):\n",
        "  for i in range(len(data)):\n",
        "    data[i].append(target[i])\n",
        "  return pd.DataFrame(data)        "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-0bBs9g3RBM3",
        "colab_type": "code",
        "outputId": "40731865-cddc-4452-a049-04453a956835",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 399
        }
      },
      "source": [
        "df_copy=getHurst(df_copy)\n",
        "getStatsForHurst(df_copy)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "mean value of hurst exponent for class 1: 0.17426759520249896\n",
            "mean value of hurst exponent for class 0: 0.5044276844723695\n",
            "mean value of hurst constant for class 1: 22.537759589041105\n",
            "mean value of hurst constant for class 0: 9.76579181307516\n",
            "median value of hurst exponent for class 1: 0.16788038442091022\n",
            "median value of hurst exponent for class 0: 0.5082046350724865\n",
            "median value of hurst constant for class 1: 10.294529749768916\n",
            "median value of hurst constant for class 0: 5.251374799982095\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u1uPdvj3PpjH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df_copy_fea=getWaveletFeatures(df_copy,target)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zjrpbB_pRuzu",
        "colab_type": "code",
        "outputId": "3bc678e1-d5e8-4725-dee3-bab0f472eb10",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 215
        }
      },
      "source": [
        "df_copy_fea.head()#our new dataset is ready"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>0</th>\n",
              "      <th>1</th>\n",
              "      <th>2</th>\n",
              "      <th>3</th>\n",
              "      <th>4</th>\n",
              "      <th>5</th>\n",
              "      <th>6</th>\n",
              "      <th>7</th>\n",
              "      <th>8</th>\n",
              "      <th>9</th>\n",
              "      <th>10</th>\n",
              "      <th>11</th>\n",
              "      <th>12</th>\n",
              "      <th>13</th>\n",
              "      <th>14</th>\n",
              "      <th>15</th>\n",
              "      <th>16</th>\n",
              "      <th>17</th>\n",
              "      <th>18</th>\n",
              "      <th>19</th>\n",
              "      <th>20</th>\n",
              "      <th>21</th>\n",
              "      <th>22</th>\n",
              "      <th>23</th>\n",
              "      <th>24</th>\n",
              "      <th>25</th>\n",
              "      <th>26</th>\n",
              "      <th>27</th>\n",
              "      <th>28</th>\n",
              "      <th>29</th>\n",
              "      <th>30</th>\n",
              "      <th>31</th>\n",
              "      <th>32</th>\n",
              "      <th>33</th>\n",
              "      <th>34</th>\n",
              "      <th>35</th>\n",
              "      <th>36</th>\n",
              "      <th>37</th>\n",
              "      <th>38</th>\n",
              "      <th>39</th>\n",
              "      <th>40</th>\n",
              "      <th>41</th>\n",
              "      <th>42</th>\n",
              "      <th>43</th>\n",
              "      <th>44</th>\n",
              "      <th>45</th>\n",
              "      <th>46</th>\n",
              "      <th>47</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.546585</td>\n",
              "      <td>2.601079</td>\n",
              "      <td>-305.176164</td>\n",
              "      <td>-151.216527</td>\n",
              "      <td>723.449162</td>\n",
              "      <td>776.917001</td>\n",
              "      <td>-3.295061</td>\n",
              "      <td>185.214508</td>\n",
              "      <td>421.056252</td>\n",
              "      <td>1.772884e+05</td>\n",
              "      <td>362.384596</td>\n",
              "      <td>-118.255497</td>\n",
              "      <td>-14.756858</td>\n",
              "      <td>112.742925</td>\n",
              "      <td>181.068760</td>\n",
              "      <td>38.918875</td>\n",
              "      <td>43.017619</td>\n",
              "      <td>96.291006</td>\n",
              "      <td>9271.957908</td>\n",
              "      <td>82.299186</td>\n",
              "      <td>-282.140173</td>\n",
              "      <td>-93.356560</td>\n",
              "      <td>83.190719</td>\n",
              "      <td>241.024738</td>\n",
              "      <td>-19.130124</td>\n",
              "      <td>-15.012851</td>\n",
              "      <td>159.427454</td>\n",
              "      <td>25417.113146</td>\n",
              "      <td>127.616292</td>\n",
              "      <td>-84.288361</td>\n",
              "      <td>-27.085687</td>\n",
              "      <td>25.839238</td>\n",
              "      <td>64.083271</td>\n",
              "      <td>-5.246926</td>\n",
              "      <td>-2.354776</td>\n",
              "      <td>46.331585</td>\n",
              "      <td>2146.615760</td>\n",
              "      <td>36.145883</td>\n",
              "      <td>-14.957938</td>\n",
              "      <td>-5.592195</td>\n",
              "      <td>5.819614</td>\n",
              "      <td>13.353548</td>\n",
              "      <td>0.068490</td>\n",
              "      <td>-0.211866</td>\n",
              "      <td>9.135993</td>\n",
              "      <td>83.466361</td>\n",
              "      <td>7.263468</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>-0.254452</td>\n",
              "      <td>75.697950</td>\n",
              "      <td>-1242.993850</td>\n",
              "      <td>-488.612159</td>\n",
              "      <td>1410.559480</td>\n",
              "      <td>1774.820173</td>\n",
              "      <td>603.491844</td>\n",
              "      <td>509.288143</td>\n",
              "      <td>1217.249973</td>\n",
              "      <td>1.481697e+06</td>\n",
              "      <td>1139.870482</td>\n",
              "      <td>-1513.818103</td>\n",
              "      <td>-506.059703</td>\n",
              "      <td>178.143757</td>\n",
              "      <td>1402.623818</td>\n",
              "      <td>-40.415845</td>\n",
              "      <td>-140.425072</td>\n",
              "      <td>816.324380</td>\n",
              "      <td>666385.493673</td>\n",
              "      <td>567.598282</td>\n",
              "      <td>-723.611943</td>\n",
              "      <td>-180.580005</td>\n",
              "      <td>125.722567</td>\n",
              "      <td>867.058241</td>\n",
              "      <td>-18.439261</td>\n",
              "      <td>-27.032981</td>\n",
              "      <td>483.947890</td>\n",
              "      <td>234205.560224</td>\n",
              "      <td>313.691461</td>\n",
              "      <td>-230.912721</td>\n",
              "      <td>-40.390111</td>\n",
              "      <td>40.967077</td>\n",
              "      <td>203.621093</td>\n",
              "      <td>0.179625</td>\n",
              "      <td>-1.976516</td>\n",
              "      <td>213.647453</td>\n",
              "      <td>45645.234141</td>\n",
              "      <td>105.978403</td>\n",
              "      <td>-24.898853</td>\n",
              "      <td>-6.318527</td>\n",
              "      <td>4.878626</td>\n",
              "      <td>33.563928</td>\n",
              "      <td>-0.514451</td>\n",
              "      <td>-0.682946</td>\n",
              "      <td>30.317390</td>\n",
              "      <td>919.144152</td>\n",
              "      <td>14.445416</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.742152</td>\n",
              "      <td>3.615098</td>\n",
              "      <td>-309.103358</td>\n",
              "      <td>-210.090841</td>\n",
              "      <td>-117.128242</td>\n",
              "      <td>-3.579475</td>\n",
              "      <td>-153.466544</td>\n",
              "      <td>-160.911085</td>\n",
              "      <td>93.143353</td>\n",
              "      <td>8.675684e+03</td>\n",
              "      <td>160.911085</td>\n",
              "      <td>-194.163602</td>\n",
              "      <td>-44.434483</td>\n",
              "      <td>2.659560</td>\n",
              "      <td>35.806625</td>\n",
              "      <td>-11.508281</td>\n",
              "      <td>-31.405705</td>\n",
              "      <td>67.695044</td>\n",
              "      <td>4582.618989</td>\n",
              "      <td>44.044606</td>\n",
              "      <td>-110.870327</td>\n",
              "      <td>-48.401351</td>\n",
              "      <td>48.215553</td>\n",
              "      <td>90.924135</td>\n",
              "      <td>5.788707</td>\n",
              "      <td>-0.261057</td>\n",
              "      <td>66.397043</td>\n",
              "      <td>4408.567255</td>\n",
              "      <td>53.872532</td>\n",
              "      <td>-26.976553</td>\n",
              "      <td>-17.167687</td>\n",
              "      <td>16.434903</td>\n",
              "      <td>29.220142</td>\n",
              "      <td>2.497897</td>\n",
              "      <td>0.764631</td>\n",
              "      <td>19.968472</td>\n",
              "      <td>398.739870</td>\n",
              "      <td>17.381640</td>\n",
              "      <td>-7.360370</td>\n",
              "      <td>-3.528846</td>\n",
              "      <td>3.431295</td>\n",
              "      <td>6.996356</td>\n",
              "      <td>-0.295030</td>\n",
              "      <td>-0.251944</td>\n",
              "      <td>4.958808</td>\n",
              "      <td>24.589778</td>\n",
              "      <td>3.869186</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.767329</td>\n",
              "      <td>12.232493</td>\n",
              "      <td>-392.667105</td>\n",
              "      <td>-384.158663</td>\n",
              "      <td>-274.167936</td>\n",
              "      <td>-168.574097</td>\n",
              "      <td>-295.871642</td>\n",
              "      <td>-294.446636</td>\n",
              "      <td>96.600146</td>\n",
              "      <td>9.331588e+03</td>\n",
              "      <td>295.594845</td>\n",
              "      <td>-50.378062</td>\n",
              "      <td>-6.343640</td>\n",
              "      <td>22.221957</td>\n",
              "      <td>56.380391</td>\n",
              "      <td>8.802419</td>\n",
              "      <td>8.508557</td>\n",
              "      <td>33.808497</td>\n",
              "      <td>1143.014456</td>\n",
              "      <td>26.964985</td>\n",
              "      <td>-49.878219</td>\n",
              "      <td>-17.150750</td>\n",
              "      <td>19.639245</td>\n",
              "      <td>34.120662</td>\n",
              "      <td>7.996985</td>\n",
              "      <td>2.005385</td>\n",
              "      <td>26.269508</td>\n",
              "      <td>690.087074</td>\n",
              "      <td>21.719040</td>\n",
              "      <td>-13.314688</td>\n",
              "      <td>-5.828000</td>\n",
              "      <td>5.282980</td>\n",
              "      <td>14.428778</td>\n",
              "      <td>-0.699589</td>\n",
              "      <td>-0.378611</td>\n",
              "      <td>11.801291</td>\n",
              "      <td>139.270476</td>\n",
              "      <td>8.033911</td>\n",
              "      <td>-5.068773</td>\n",
              "      <td>-2.264661</td>\n",
              "      <td>2.383344</td>\n",
              "      <td>5.377493</td>\n",
              "      <td>0.336105</td>\n",
              "      <td>0.040430</td>\n",
              "      <td>5.218204</td>\n",
              "      <td>27.229656</td>\n",
              "      <td>3.162152</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.647870</td>\n",
              "      <td>2.538971</td>\n",
              "      <td>-276.585954</td>\n",
              "      <td>-255.283369</td>\n",
              "      <td>5.788142</td>\n",
              "      <td>87.801277</td>\n",
              "      <td>-79.855665</td>\n",
              "      <td>-100.024326</td>\n",
              "      <td>137.229131</td>\n",
              "      <td>1.883183e+04</td>\n",
              "      <td>137.196686</td>\n",
              "      <td>-118.472221</td>\n",
              "      <td>-50.295091</td>\n",
              "      <td>15.728461</td>\n",
              "      <td>61.397235</td>\n",
              "      <td>-13.467008</td>\n",
              "      <td>-16.719396</td>\n",
              "      <td>59.952932</td>\n",
              "      <td>3594.354091</td>\n",
              "      <td>49.218659</td>\n",
              "      <td>-91.214554</td>\n",
              "      <td>-51.554051</td>\n",
              "      <td>31.027079</td>\n",
              "      <td>83.607092</td>\n",
              "      <td>2.235106</td>\n",
              "      <td>-5.263506</td>\n",
              "      <td>53.373437</td>\n",
              "      <td>2848.723725</td>\n",
              "      <td>44.339743</td>\n",
              "      <td>-47.796112</td>\n",
              "      <td>-13.439282</td>\n",
              "      <td>21.991572</td>\n",
              "      <td>49.627529</td>\n",
              "      <td>0.224143</td>\n",
              "      <td>1.325679</td>\n",
              "      <td>29.940156</td>\n",
              "      <td>896.412956</td>\n",
              "      <td>22.968023</td>\n",
              "      <td>-9.831802</td>\n",
              "      <td>-3.274258</td>\n",
              "      <td>3.681270</td>\n",
              "      <td>9.592965</td>\n",
              "      <td>-0.133232</td>\n",
              "      <td>0.179974</td>\n",
              "      <td>7.101117</td>\n",
              "      <td>50.425867</td>\n",
              "      <td>4.900663</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "         0          1            2   ...          45         46  47\n",
              "0  0.546585   2.601079  -305.176164  ...   83.466361   7.263468   0\n",
              "1 -0.254452  75.697950 -1242.993850  ...  919.144152  14.445416   1\n",
              "2  0.742152   3.615098  -309.103358  ...   24.589778   3.869186   0\n",
              "3  0.767329  12.232493  -392.667105  ...   27.229656   3.162152   0\n",
              "4  0.647870   2.538971  -276.585954  ...   50.425867   4.900663   0\n",
              "\n",
              "[5 rows x 48 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3b4eLANfSaoc",
        "colab_type": "text"
      },
      "source": [
        "**Resampling data**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ecd2dHmKdjQb",
        "colab_type": "text"
      },
      "source": [
        "Before the resampling, the 0 class was too prominent.I tried to balance the data by shuffling and resampling the data, In order get a more balanced dataset for machine learning models."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tDkbngBsYGnj",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.utils import shuffle\n",
        "def createBalancedDataset(data,random_state):\n",
        "  #shuffling for random sampling\n",
        "  X = shuffle(data,random_state=random_state)\n",
        "  #getting first 6500 value\n",
        "  return X.sort_values(by=47, ascending=False).iloc[:6500].index"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ysXu9dXnYvr3",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "v=createBalancedDataset(df_copy_fea,42)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Vi95z9uNYl4x",
        "colab_type": "code",
        "outputId": "e12baf34-1f87-4b9e-e90c-d8a69c9988ca",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 315
        }
      },
      "source": [
        "plt.hist((df_copy_fea.iloc[v])[47])\n",
        "(df_copy_fea.iloc[v][47]).value_counts() #more balanced dataset"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0    4200\n",
              "1    2300\n",
              "Name: 47, dtype: int64"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qVpNzLGhcAfV",
        "colab_type": "text"
      },
      "source": [
        "**Normalizing Data**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dzxo07gNcEH_",
        "colab_type": "text"
      },
      "source": [
        "Before the feature selection I normalized data.Normalizing data is necessary for Anova test.Also normalizing is a prerequisite for a lot of machine learning algorithms[11]."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EWI2ZK9iSZtV",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#normalizing dataset\n",
        "from sklearn.preprocessing import MinMaxScaler\n",
        "scaler = MinMaxScaler()\n",
        "scaler.fit(df_copy_fea.drop([47],axis=1))\n",
        "n_df_fea=pd.DataFrame(scaler.transform(df_copy_fea.drop([47],axis=1)))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TVN3LLg7fYHP",
        "colab_type": "text"
      },
      "source": [
        "**Feature Selection with Anova and Random Forest**"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JDeGkYPGgW9L",
        "colab_type": "text"
      },
      "source": [
        "Before giving inputs to models.I inspected whether feature selection improve the accuracies.In order to understand this situation, first i gave all features to model and then i gave only the selected features as input to model and i compared the accuries and f-scores.\n",
        "\n",
        "For feature selection, I got help by below diagram.Since the dataset's input values are numeric but target variable is categoric, I choosed the Anova test for feature selection.Furthermore i did one more feature selection with random forest in order to compare them together.Eventhough feature selection can be very helpful on improving accuracies in many case, In this situation, it did not help so I decided to continue with all features as input.The process is below.\n",
        "\n",
        "![alt text](https://3qeqpr26caki16dnhd19sv6by6v-wpengine.netdna-ssl.com/wp-content/uploads/2019/11/How-to-Choose-Feature-Selection-Methods-For-Machine-Learning.png)\n",
        "\n",
        "(picture is taken from [12])\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pNuG6tJBktP_",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#First the all features are given as a input to SVM model (I will explain why i used SVM later)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LlFLwQmdZUks",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "X_trainr, X_testr, y_trainr, y_testr = train_test_split(n_df_fea.iloc[v], target.iloc[v], test_size=0.33, random_state=42)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "b5AVQeNklxSc",
        "colab_type": "code",
        "outputId": "eeeec60a-ca66-4b66-fa7d-d200b8344bc9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        }
      },
      "source": [
        "#Reference point for svm\n",
        "from sklearn import svm\n",
        "from sklearn.model_selection import cross_val_predict\n",
        "from sklearn.metrics import classification_report\n",
        "\n",
        "#I will explain this model in model part in the notebook\n",
        "clf = svm.SVC(kernel=\"linear\")\n",
        "clf.fit(X_trainr, y_trainr)\n",
        "#cross validation is 10\n",
        "y_pred = cross_val_predict(clf,X_testr,y_testr,cv=10)\n",
        "print(\"All features are inclueded\\n\",classification_report(y_testr, y_pred))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "All features are inclueded\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0       0.96      0.99      0.97      1373\n",
            "           1       0.98      0.93      0.95       772\n",
            "\n",
            "    accuracy                           0.97      2145\n",
            "   macro avg       0.97      0.96      0.96      2145\n",
            "weighted avg       0.97      0.97      0.97      2145\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4hp45deKqOZs",
        "colab_type": "text"
      },
      "source": [
        "**Selecting most important 20 features with Anova**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X4lxkNVepcWH",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.feature_selection import SelectKBest\n",
        "from sklearn.feature_selection import f_classif\n",
        "#Selection most important 20 feature by using Anova test\n",
        "def selectFeature(X_trainr,y_trainr,X_testr):\n",
        "  sel_f = SelectKBest(f_classif, k=20)\n",
        "  X_train_f = sel_f.fit_transform(X_trainr, y_trainr)\n",
        "  mySelectedFeatures=[i for i in range(len(sel_f.get_support())) if sel_f.get_support()[i]==True]\n",
        "  j=0\n",
        "  unseable_columns=[]\n",
        "  #Creating a new dataset with these 20 features\n",
        "  for i in X_trainr.columns:\n",
        "    if(j not in mySelectedFeatures):\n",
        "      unseable_columns.append(i)\n",
        "    j+=1 \n",
        "  X_train_arranged=X_trainr.drop(columns=unseable_columns)\n",
        "  X_test_arranged=X_testr.drop(columns=unseable_columns)\n",
        "  return  X_train_arranged,X_test_arranged\n",
        "\n",
        "X_train_arranged,X_test_arranged=selectFeature(X_trainr,y_trainr,X_testr)  "
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "sXNdWS2XyUFS",
        "colab_type": "code",
        "outputId": "e986ae84-7744-4717-d9f1-66825750077c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 67
        }
      },
      "source": [
        "X_train_arranged.columns #The most important columns according to Anova"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Int64Index([2, 5, 8, 10, 11, 14, 17, 19, 20, 21, 22, 23, 26, 28, 29, 32, 35,\n",
              "            37, 44, 46],\n",
              "           dtype='int64')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "42gOSwhLpSdC",
        "colab_type": "code",
        "outputId": "4fcd660a-2259-4333-b445-e29dc1065af4",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        }
      },
      "source": [
        "#Overall accuracy is decreased\n",
        "from sklearn import svm\n",
        "from sklearn.metrics import classification_report\n",
        "clf = svm.SVC(kernel=\"linear\")\n",
        "clf.fit(X_train_arranged, y_trainr)\n",
        "y_pred = cross_val_predict(clf,X_test_arranged,y_testr,cv=10)\n",
        "print(\"Only Anova test's Features are used\\n\",classification_report(y_testr, y_pred))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Only Anova test's Features are used\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0       0.96      0.98      0.97      1373\n",
            "           1       0.97      0.92      0.95       772\n",
            "\n",
            "    accuracy                           0.96      2145\n",
            "   macro avg       0.96      0.95      0.96      2145\n",
            "weighted avg       0.96      0.96      0.96      2145\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AXGoLK1EscE_",
        "colab_type": "text"
      },
      "source": [
        "**Selecting most important 20 features with Random Forest**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XIpLs6yLroJ-",
        "colab_type": "code",
        "outputId": "8535e59f-4624-454b-c4b2-711347ee41ac",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "#Firstly I used grid Search for getting best hyperparameter for random-forest\n",
        "np.random.seed(42)\n",
        "from sklearn.model_selection import GridSearchCV\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "rfc = RandomForestClassifier(n_jobs=-1,max_features= 'sqrt' ,n_estimators=50, oob_score = True) \n",
        "\n",
        "param_grid = { \n",
        "    'max_depth': [2,5],\n",
        "    'min_samples_split':[2,5,10],\n",
        "    'n_estimators': [100,150],\n",
        "    'max_features': ['auto', 'sqrt', 'log2']\n",
        "}\n",
        "\n",
        "CV_rfc = GridSearchCV(estimator=rfc, param_grid=param_grid, cv=5)\n",
        "CV_rfc.fit(X_trainr, y_trainr)\n",
        "print (CV_rfc.best_params_)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "{'max_depth': 5, 'max_features': 'sqrt', 'min_samples_split': 2, 'n_estimators': 150}\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "RsVw0rELwOiA",
        "colab_type": "code",
        "outputId": "02caa3fd-e01c-4792-bd61-0674caac7375",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        }
      },
      "source": [
        "from sklearn.ensemble import RandomForestClassifier\n",
        "clf = RandomForestClassifier(random_state=42,max_depth=5,max_features='sqrt',min_samples_split=2,n_estimators=150)\n",
        "clf.fit(X_trainr, y_trainr)\n",
        "#I also get the importance rates and sort in a desending order and create a dataframe\n",
        "zipped=pd.DataFrame(zip(X_trainr.columns,clf.feature_importances_),columns=[\"column\",\"importance\"]).sort_values(by=\"importance\", ascending=False)\n",
        "y_pred2 = cross_val_predict(clf,X_testr,y_testr,cv=10)\n",
        "print(\"All featuares are included\\n\",classification_report(y_testr, y_pred2))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "All featuares are included\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      0.98      0.98      1373\n",
            "           1       0.96      0.96      0.96       772\n",
            "\n",
            "    accuracy                           0.97      2145\n",
            "   macro avg       0.97      0.97      0.97      2145\n",
            "weighted avg       0.97      0.97      0.97      2145\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "64thlIPiyC5x",
        "colab_type": "code",
        "outputId": "5e5c30ae-bb7e-42fb-9242-6c12b9ee1c6a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 639
        }
      },
      "source": [
        "zipped.head(20) #The most important 20 values according to random forest"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>column</th>\n",
              "      <th>importance</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>18</td>\n",
              "      <td>0.122870</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>17</td>\n",
              "      <td>0.102400</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>8</td>\n",
              "      <td>0.086825</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>9</td>\n",
              "      <td>0.083827</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>19</td>\n",
              "      <td>0.082742</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>36</th>\n",
              "      <td>36</td>\n",
              "      <td>0.075209</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>35</th>\n",
              "      <td>35</td>\n",
              "      <td>0.065976</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>14</td>\n",
              "      <td>0.047009</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>11</td>\n",
              "      <td>0.045067</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>26</th>\n",
              "      <td>26</td>\n",
              "      <td>0.036911</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>29</th>\n",
              "      <td>29</td>\n",
              "      <td>0.035794</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>32</th>\n",
              "      <td>32</td>\n",
              "      <td>0.032185</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>20</td>\n",
              "      <td>0.024624</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>28</th>\n",
              "      <td>28</td>\n",
              "      <td>0.017470</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>10</td>\n",
              "      <td>0.016753</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>27</th>\n",
              "      <td>27</td>\n",
              "      <td>0.015615</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>23</th>\n",
              "      <td>23</td>\n",
              "      <td>0.012727</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>37</th>\n",
              "      <td>37</td>\n",
              "      <td>0.012705</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>2</td>\n",
              "      <td>0.011719</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>13</td>\n",
              "      <td>0.009039</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "    column  importance\n",
              "18      18    0.122870\n",
              "17      17    0.102400\n",
              "8        8    0.086825\n",
              "9        9    0.083827\n",
              "19      19    0.082742\n",
              "36      36    0.075209\n",
              "35      35    0.065976\n",
              "14      14    0.047009\n",
              "11      11    0.045067\n",
              "26      26    0.036911\n",
              "29      29    0.035794\n",
              "32      32    0.032185\n",
              "20      20    0.024624\n",
              "28      28    0.017470\n",
              "10      10    0.016753\n",
              "27      27    0.015615\n",
              "23      23    0.012727\n",
              "37      37    0.012705\n",
              "2        2    0.011719\n",
              "13      13    0.009039"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 29
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hspKn5NwlaGG",
        "colab_type": "code",
        "outputId": "2314934b-6b24-44c3-c7c1-099fd2e72300",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        }
      },
      "source": [
        "#F-scores are decreased\n",
        "clf = svm.SVC(kernel=\"linear\")\n",
        "clf.fit(X_trainr[zipped.iloc[:20].index], y_trainr)\n",
        "y_pred = cross_val_predict(clf,X_testr[zipped.iloc[:20].index],y_testr,cv=10)\n",
        "print(\"Only random forest's features are inclueded\\n\",classification_report(y_testr, y_pred))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Only random forest's features are inclueded\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0       0.96      0.98      0.97      1373\n",
            "           1       0.96      0.92      0.94       772\n",
            "\n",
            "    accuracy                           0.96      2145\n",
            "   macro avg       0.96      0.95      0.96      2145\n",
            "weighted avg       0.96      0.96      0.96      2145\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qLH2LYdZDgJS",
        "colab_type": "text"
      },
      "source": [
        "![alt text](https://drive.google.com/uc?id=1VwWDqamqvC8_W0CHLsWfKYLmhJVmPmKM)\n",
        "\n",
        "As you can see, The feature selection process did not increase accuracies.Therefore I decided to continue with all features as input\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "AOJc7Aj-DgJb",
        "colab_type": "text"
      },
      "source": [
        "## Modeling<a class=\"anchor\" id=\"modeling\"></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z4v1s3Z62cyC",
        "colab_type": "text"
      },
      "source": [
        "In the model selection process, I analysed the most used algoriths in the literature.I encounter that the most used algorithms are ANNs/RNNs, SVMs and Desion Trees.Therefore I decided to use SVM since this task is a binary classification problem , Random Forest for desion trees are pretty popular in the literature and lastly LSTM for RNN's are pretty popular in the literature and it is one of my favorites."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q898U1hEDgJc",
        "colab_type": "code",
        "outputId": "57907901-94b3-4d58-a19b-6f2a6ce82553",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        }
      },
      "source": [
        "#SVM, for kernel, I used some kernels and get the most accurate one\n",
        "clf = svm.SVC(kernel=\"linear\",probability=True)\n",
        "clf.fit(X_trainr, y_trainr)\n",
        "#cross validation is 10\n",
        "y_pred = cross_val_predict(clf,X_testr,y_testr,cv=10)\n",
        "print(\"All features are included\\n\",classification_report(y_testr, y_pred))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "All features are included\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0       0.96      0.99      0.97      1373\n",
            "           1       0.98      0.93      0.95       772\n",
            "\n",
            "    accuracy                           0.97      2145\n",
            "   macro avg       0.97      0.96      0.96      2145\n",
            "weighted avg       0.97      0.97      0.97      2145\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gtUfnM-T6DCo",
        "colab_type": "code",
        "outputId": "cf1b574c-cf96-41ff-bc58-ae679b6b28ca",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        }
      },
      "source": [
        "#Random forest, I got hyperparameters from above grid-search\n",
        "clf1 = RandomForestClassifier(random_state=42,max_depth=5,max_features='auto',min_samples_split=5,n_estimators=150)\n",
        "clf1.fit(X_trainr, y_trainr)\n",
        "y_pred2 = cross_val_predict(clf1,X_testr,y_testr,cv=10)\n",
        "print(\"All featuares are included\\n\",classification_report(y_testr, y_pred2))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "All featuares are included\n",
            "               precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      0.98      0.98      1373\n",
            "           1       0.96      0.96      0.96       772\n",
            "\n",
            "    accuracy                           0.97      2145\n",
            "   macro avg       0.97      0.97      0.97      2145\n",
            "weighted avg       0.97      0.97      0.97      2145\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3lwn-vKI6L4J",
        "colab_type": "code",
        "outputId": "ec33106d-df0d-4cd4-8810-21b0b1fc716a",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "#LSTM\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Dense, Dropout\n",
        "from keras.layers import LSTM\n",
        "# When i designed the network, Basically i used heuristic approach\n",
        "X_trainrr=np.array(X_trainr).reshape(X_trainr.shape[0],X_trainr.shape[1],1)\n",
        "X_testrr=np.array(X_testr).reshape(X_testr.shape[0],X_testr.shape[1],1)\n",
        "model = Sequential()\n",
        "model.add(LSTM(50, input_shape=(X_trainrr.shape[1], X_trainrr.shape[2])))\n",
        "model.add(Dropout(0.1))\n",
        "model.add(Dense(1, activation='sigmoid'))\n",
        "\n",
        "model.compile(loss='binary_crossentropy',\n",
        "              optimizer='rmsprop',\n",
        "              metrics=['accuracy'])\n",
        "# fit network\n",
        "history = model.fit(X_trainrr, y_trainr, epochs=50, batch_size=72, validation_data=(X_testrr, y_testr), verbose=2, shuffle=False)\n",
        "# plot history\n",
        "plt.plot(history.history['loss'], label='train')\n",
        "plt.plot(history.history['val_loss'], label='test')\n",
        "plt.legend()\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Train on 4355 samples, validate on 2145 samples\n",
            "Epoch 1/50\n",
            " - 3s - loss: 0.6462 - accuracy: 0.6512 - val_loss: 0.8925 - val_accuracy: 0.6401\n",
            "Epoch 2/50\n",
            " - 2s - loss: 0.5971 - accuracy: 0.7031 - val_loss: 0.6156 - val_accuracy: 0.6606\n",
            "Epoch 3/50\n",
            " - 2s - loss: 0.5679 - accuracy: 0.7504 - val_loss: 0.6364 - val_accuracy: 0.6578\n",
            "Epoch 4/50\n",
            " - 2s - loss: 0.5287 - accuracy: 0.7451 - val_loss: 0.6908 - val_accuracy: 0.6620\n",
            "Epoch 5/50\n",
            " - 2s - loss: 0.5160 - accuracy: 0.7784 - val_loss: 0.6146 - val_accuracy: 0.6834\n",
            "Epoch 6/50\n",
            " - 2s - loss: 0.4950 - accuracy: 0.7713 - val_loss: 0.5547 - val_accuracy: 0.7324\n",
            "Epoch 7/50\n",
            " - 2s - loss: 0.4644 - accuracy: 0.8101 - val_loss: 0.5006 - val_accuracy: 0.7459\n",
            "Epoch 8/50\n",
            " - 2s - loss: 0.4782 - accuracy: 0.8051 - val_loss: 0.4023 - val_accuracy: 0.8191\n",
            "Epoch 9/50\n",
            " - 2s - loss: 0.4100 - accuracy: 0.8439 - val_loss: 0.4037 - val_accuracy: 0.8168\n",
            "Epoch 10/50\n",
            " - 2s - loss: 0.3901 - accuracy: 0.8441 - val_loss: 0.3197 - val_accuracy: 0.8625\n",
            "Epoch 11/50\n",
            " - 2s - loss: 0.4019 - accuracy: 0.8358 - val_loss: 0.3576 - val_accuracy: 0.8284\n",
            "Epoch 12/50\n",
            " - 2s - loss: 0.3585 - accuracy: 0.8641 - val_loss: 0.6342 - val_accuracy: 0.7883\n",
            "Epoch 13/50\n",
            " - 2s - loss: 0.3168 - accuracy: 0.8696 - val_loss: 0.4682 - val_accuracy: 0.8056\n",
            "Epoch 14/50\n",
            " - 2s - loss: 0.2933 - accuracy: 0.8935 - val_loss: 0.5355 - val_accuracy: 0.7977\n",
            "Epoch 15/50\n",
            " - 2s - loss: 0.3174 - accuracy: 0.8794 - val_loss: 0.2612 - val_accuracy: 0.8998\n",
            "Epoch 16/50\n",
            " - 2s - loss: 0.2807 - accuracy: 0.8909 - val_loss: 1.1429 - val_accuracy: 0.7706\n",
            "Epoch 17/50\n",
            " - 2s - loss: 0.2825 - accuracy: 0.8964 - val_loss: 0.5226 - val_accuracy: 0.8266\n",
            "Epoch 18/50\n",
            " - 2s - loss: 0.2675 - accuracy: 0.9001 - val_loss: 0.7183 - val_accuracy: 0.8298\n",
            "Epoch 19/50\n",
            " - 2s - loss: 0.2696 - accuracy: 0.8999 - val_loss: 0.6568 - val_accuracy: 0.8284\n",
            "Epoch 20/50\n",
            " - 2s - loss: 0.2497 - accuracy: 0.9077 - val_loss: 0.3833 - val_accuracy: 0.8639\n",
            "Epoch 21/50\n",
            " - 2s - loss: 0.2606 - accuracy: 0.9052 - val_loss: 0.4698 - val_accuracy: 0.8163\n",
            "Epoch 22/50\n",
            " - 2s - loss: 0.2469 - accuracy: 0.9063 - val_loss: 0.4334 - val_accuracy: 0.8643\n",
            "Epoch 23/50\n",
            " - 2s - loss: 0.2378 - accuracy: 0.9109 - val_loss: 0.3177 - val_accuracy: 0.8895\n",
            "Epoch 24/50\n",
            " - 2s - loss: 0.2349 - accuracy: 0.9116 - val_loss: 0.3822 - val_accuracy: 0.8587\n",
            "Epoch 25/50\n",
            " - 2s - loss: 0.2238 - accuracy: 0.9180 - val_loss: 0.3616 - val_accuracy: 0.8755\n",
            "Epoch 26/50\n",
            " - 2s - loss: 0.2170 - accuracy: 0.9226 - val_loss: 0.4299 - val_accuracy: 0.8639\n",
            "Epoch 27/50\n",
            " - 2s - loss: 0.2265 - accuracy: 0.9139 - val_loss: 0.3308 - val_accuracy: 0.8909\n",
            "Epoch 28/50\n",
            " - 2s - loss: 0.2199 - accuracy: 0.9146 - val_loss: 0.3164 - val_accuracy: 0.8839\n",
            "Epoch 29/50\n",
            " - 2s - loss: 0.2153 - accuracy: 0.9217 - val_loss: 0.4091 - val_accuracy: 0.8690\n",
            "Epoch 30/50\n",
            " - 2s - loss: 0.2132 - accuracy: 0.9240 - val_loss: 0.3462 - val_accuracy: 0.8765\n",
            "Epoch 31/50\n",
            " - 2s - loss: 0.2056 - accuracy: 0.9268 - val_loss: 0.3332 - val_accuracy: 0.8834\n",
            "Epoch 32/50\n",
            " - 2s - loss: 0.2062 - accuracy: 0.9263 - val_loss: 0.3056 - val_accuracy: 0.8984\n",
            "Epoch 33/50\n",
            " - 2s - loss: 0.1968 - accuracy: 0.9288 - val_loss: 0.3055 - val_accuracy: 0.8928\n",
            "Epoch 34/50\n",
            " - 2s - loss: 0.2037 - accuracy: 0.9272 - val_loss: 0.3116 - val_accuracy: 0.8946\n",
            "Epoch 35/50\n",
            " - 2s - loss: 0.1977 - accuracy: 0.9320 - val_loss: 0.3156 - val_accuracy: 0.8942\n",
            "Epoch 36/50\n",
            " - 2s - loss: 0.1971 - accuracy: 0.9290 - val_loss: 0.2872 - val_accuracy: 0.9016\n",
            "Epoch 37/50\n",
            " - 2s - loss: 0.1892 - accuracy: 0.9357 - val_loss: 0.2613 - val_accuracy: 0.9156\n",
            "Epoch 38/50\n",
            " - 2s - loss: 0.1963 - accuracy: 0.9387 - val_loss: 0.3505 - val_accuracy: 0.8900\n",
            "Epoch 39/50\n",
            " - 2s - loss: 0.1920 - accuracy: 0.9284 - val_loss: 0.4161 - val_accuracy: 0.8830\n",
            "Epoch 40/50\n",
            " - 2s - loss: 0.1862 - accuracy: 0.9332 - val_loss: 0.3528 - val_accuracy: 0.8960\n",
            "Epoch 41/50\n",
            " - 2s - loss: 0.1871 - accuracy: 0.9295 - val_loss: 0.5083 - val_accuracy: 0.8918\n",
            "Epoch 42/50\n",
            " - 2s - loss: 0.1822 - accuracy: 0.9350 - val_loss: 1.0790 - val_accuracy: 0.8350\n",
            "Epoch 43/50\n",
            " - 2s - loss: 0.1933 - accuracy: 0.9380 - val_loss: 0.2751 - val_accuracy: 0.8937\n",
            "Epoch 44/50\n",
            " - 2s - loss: 0.1776 - accuracy: 0.9371 - val_loss: 0.4427 - val_accuracy: 0.8993\n",
            "Epoch 45/50\n",
            " - 2s - loss: 0.1794 - accuracy: 0.9348 - val_loss: 0.4031 - val_accuracy: 0.8974\n",
            "Epoch 46/50\n",
            " - 2s - loss: 0.1806 - accuracy: 0.9357 - val_loss: 0.3848 - val_accuracy: 0.8657\n",
            "Epoch 47/50\n",
            " - 2s - loss: 0.1686 - accuracy: 0.9398 - val_loss: 0.4618 - val_accuracy: 0.8984\n",
            "Epoch 48/50\n",
            " - 2s - loss: 0.1798 - accuracy: 0.9352 - val_loss: 0.2655 - val_accuracy: 0.9212\n",
            "Epoch 49/50\n",
            " - 2s - loss: 0.1738 - accuracy: 0.9366 - val_loss: 0.2365 - val_accuracy: 0.9198\n",
            "Epoch 50/50\n",
            " - 2s - loss: 0.1692 - accuracy: 0.9385 - val_loss: 0.3694 - val_accuracy: 0.9058\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zDlpPDydDgJu",
        "colab_type": "text"
      },
      "source": [
        "## Evaluation<a class=\"anchor\" id=\"evaluation\"></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "iHkV_7Z9--6g",
        "colab_type": "text"
      },
      "source": [
        "For evalaute the model, I used f-scores, ROC curves and overall accuracies.I provided the models accuracies and f-scores as table below.I also provided ROC curves for SVM and Random Forest.\n",
        "\n",
        "![alt text](https://drive.google.com/uc?id=1Z53mCM5oPJ-yFUI6T1g647KUPrfCTozf)\n",
        "\n",
        "As you see the most accurate algorithm is Random Forest and all algorithms are more successful at classifiying 0 class than 1 class."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "keJzq2EpjIq0",
        "colab_type": "text"
      },
      "source": [
        "**ROC Curves**"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "js87y7dkGt__",
        "colab_type": "code",
        "outputId": "295d7755-7c16-4590-f190-20d6d875e144",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 252
        }
      },
      "source": [
        "pip install scikit-plot"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting scikit-plot\n",
            "  Downloading https://files.pythonhosted.org/packages/7c/47/32520e259340c140a4ad27c1b97050dd3254fdc517b1d59974d47037510e/scikit_plot-0.3.7-py3-none-any.whl\n",
            "Requirement already satisfied: joblib>=0.10 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (0.15.1)\n",
            "Requirement already satisfied: scikit-learn>=0.18 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (0.22.2.post1)\n",
            "Requirement already satisfied: scipy>=0.9 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (1.4.1)\n",
            "Requirement already satisfied: matplotlib>=1.4.0 in /usr/local/lib/python3.6/dist-packages (from scikit-plot) (3.2.1)\n",
            "Requirement already satisfied: numpy>=1.11.0 in /usr/local/lib/python3.6/dist-packages (from scikit-learn>=0.18->scikit-plot) (1.18.4)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (0.10.0)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (1.2.0)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (2.8.1)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib>=1.4.0->scikit-plot) (2.4.7)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from cycler>=0.10->matplotlib>=1.4.0->scikit-plot) (1.12.0)\n",
            "Installing collected packages: scikit-plot\n",
            "Successfully installed scikit-plot-0.3.7\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CJKvrYH7Gacw",
        "colab_type": "code",
        "outputId": "82a6a7ee-216a-4831-8677-4639bf7bb546",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 349
        }
      },
      "source": [
        "import scikitplot as skplt\n",
        "import matplotlib.pyplot as plt\n",
        "#ROC Curve for SVM\n",
        "y_pred_proba=clf.predict_proba(X_testr)\n",
        "skplt.metrics.plot_roc_curve(y_testr,y_pred_proba)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function plot_roc_curve is deprecated; This will be removed in v0.5.0. Please use scikitplot.metrics.plot_roc instead.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nYHOP4VIItMf",
        "colab_type": "code",
        "outputId": "fce28fc0-4e80-4249-bfee-814d00e7fbec",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 349
        }
      },
      "source": [
        "#ROC curve for Random-Forest\n",
        "y_pred_proba=clf1.predict_proba(X_testr)\n",
        "skplt.metrics.plot_roc_curve(y_testr,y_pred_proba)\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function plot_roc_curve is deprecated; This will be removed in v0.5.0. Please use scikitplot.metrics.plot_roc instead.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TBc96NX0JZd0",
        "colab_type": "text"
      },
      "source": [
        "As you can see both models ROC curve areas,f-scores of both classes and overall accuracies are very close to 1 and %100.Therefore the models are very accurate."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ZzVYPKtiDgKE",
        "colab_type": "text"
      },
      "source": [
        "## References<a class=\"anchor\" id=\"references\"></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yL3uvJ0iDgKK",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "\n",
        "$$$$\n",
        "[1] \tS. A. Hosseini, M-R. Akbarzadeh-T,M-B. Naghibi-Sistani, «Qualitative and Quantitative Evaluation of EEG,» MECS, 2013.\n",
        "\n",
        "[2] \tLing Guo, Daniel Rivero, Julián Dorado, Juan R. Rabunal, Alejandro Pazos, «Automatic epileptic seizure detection in EEGs based on line length feature and,» Journal of Neuroscience Methods, 2010. \n",
        "\n",
        "[3]Chang BS, Lowenstein DH. Epilepsy. N Engl J Med. 2003;349(13):1257‐1266. doi:10.1056/NEJMra022308\n",
        "\n",
        "[4]Eadie MJ. Shortcomings in the current treatment of epilepsy. Expert Rev Neurother. 2012;12(12):1419‐1427. doi:10.1586/ern.12.129\n",
        "\n",
        "[5]D. D. A. ÖZKARDEŞ. [Online]. Available: https://www.memorial.com.tr/saglik-rehberleri/epilepsi-ile-ilgili-bilmeniz-gereken-10-gercek/.\n",
        "\n",
        "[6]Yuan, Q., Li, F., & Zhong, H. (2015). Early diagnosis, treatment and prognosis of epilepsy with continuous spikes and waves during slow sleep. International journal of clinical and experimental medicine, 8(3), 4052–4058.\n",
        "\n",
        "[7]«Hurst Exponent,» [Online]. Available: https://en.wikipedia.org/wiki/Hurst_exponent.\n",
        "\n",
        "[8]«Discrete Wavelet Transform,» [Online]. Available: https://en.wikipedia.org/wiki/Discrete_wavelet_transform.\n",
        "\n",
        "[9]A. Taşpınar, «A guide for using the Wavelet Transform in Machine Learning,» [Online]. Available: http://ataspinar.com/2018/12/21/a-guide-for-using-the-wavelet-transform-in-machine-learning/.\n",
        "\n",
        "[10][Online]. Available: http://ataspinar.com/wp-content/uploads/2018/08/Comparisonoftransformations.jpg.\n",
        "\n",
        "[11]A. Chavakula. [Online]. Available: https://www.quora.com/Which-machine-algorithms-require-data-scaling-normalization.\n",
        "\n",
        "[12]J. Brownlee, «Machine Learning Mastery,» [Online]. Available: https://3qeqpr26caki16dnhd19sv6by6v-wpengine.netdna-ssl.com/wp-content/uploads/2019/11/How-to-Choose-Feature-Selection-Methods-For-Machine-Learning.png.\n",
        "\n",
        "[13]Faust, O., Acharya, U. R., Adeli, H., & Adeli, A. (2015). Wavelet-based EEG processing for computer-aided seizure detection and epilepsy diagnosis. Seizure, 26, 56-64.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sOrVtZb6DgKT",
        "colab_type": "text"
      },
      "source": [
        "\n",
        "**Disclaimer!** <font color='grey'>This notebook was prepared by Nur Altıparmak as a term project for the *BBM469 - Data Intensive Applications Laboratory* class. The notebook is available for educational purposes only. There is no guarantee on the correctness of the content provided as it is a student work.\n",
        "\n",
        "If you think there is any copyright violation, please let us [know](https://forms.gle/BNNRB2kR8ZHVEREq8). \n",
        "</font>"
      ]
    }
  ]
}