cap2_2_feature_extraction_sift.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "cap2_2_feature_extraction_sift.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyOUykR7pbtK1v1yxq11EGU7",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/viniciusrpb/cic0193_machinelearning/blob/main/cap2_2_feature_extraction_sift.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "zZqsThx2qSPo"
      },
      "source": [
        "# Capítulo 2\n",
        "\n",
        "## 2.2. Extração de Características de Imagens utilizando Bag-of-Features (Bag-of-visual-words) com Scale Invariant Feature Transform (SIFT)\n",
        "\n",
        "Vamos utilizar a coleção de imagens Swedish Leaf Dataset disponível aqui:\n",
        "\n",
        "https://www.cvl.isy.liu.se/en/research/datasets/swedish-leaf/\n",
        "\n",
        "Particularmente, será utilizada a classe \"1. Ulmus carpinifolia\" e todas as 75 imagens foram convertidas de \".tif\" para \".png\" antes da execução desse notebook.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0efJ36bh6ZR-"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount(\"/content/drive\",force_remount=True)"
      ],
      "execution_count": 24,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qQpUMjXpK4y7"
      },
      "source": [
        "!cp -r \"/content/drive/My Drive/leafs\" \"leafs\""
      ],
      "execution_count": 25,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tvxVYp7Yp-gk"
      },
      "source": [
        "Como a implementação do SIFT não está disponível na versão mais atual do opencv devido a uma patente, é preciso instalar uma versão mais antiga do opencv que tinha permissão para ter essa implementação"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S4ysBRyFLCM2"
      },
      "source": [
        "!pip3 install opencv-python==3.4.2.17\n",
        "!pip3 install opencv-contrib-python==3.4.2.17"
      ],
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Ab6l9XMqrNW2"
      },
      "source": [
        "Importando as bibliotecas necessárias\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cQrFjLz3LbUh"
      },
      "source": [
        "import numpy as np\n",
        "import cv2\n",
        "import os\n",
        "from os import listdir\n",
        "from sklearn.cluster import KMeans\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt"
      ],
      "execution_count": 27,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9jrUHELNrT11"
      },
      "source": [
        "Pega o nome dos arquivos associados às imagens das folhas"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XCfDWsvpL9OQ"
      },
      "source": [
        "path_imgs = \"leafs/leaf1/\"\n",
        "\n",
        "l_imgs = listdir(path_imgs)"
      ],
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vuU0-GpWrdbs"
      },
      "source": [
        "## Implementação do Bag-of-Features\n",
        "\n",
        "Primeiramente, deve-se inicializar uma lista relacionada a um dicionário, que vai armazenar os descritores de todos os \"patches\" (localidades) das imagens da entrada.\n",
        "\n",
        "Em seguida, executa-se o SIFT, que retorna os keypoints (pontos locais) e o histograma das orientações para cada imagem da entrada. É importante salientar que cada imagem pode gerar um número diferentes de pontos-chave e o descritor sempre possui 128 posições.\n",
        "\n",
        "Vale ressaltar que cada patch está associado com a região de um keypoint. Assim, o descritor de cada keypoint vai representar um vetor de características."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lKOhEKjcMSdM"
      },
      "source": [
        "dictionary = []\n",
        "\n",
        "for img_path in l_imgs:\n",
        "\n",
        "  img = cv2.imread(path_imgs+img_path)\n",
        "\n",
        "  sift = cv2.xfeatures2d.SIFT_create()\n",
        "\n",
        "  keypoints,descriptor = sift.detectAndCompute(img,None)\n",
        "\n",
        "  for d in descriptor:\n",
        "    dictionary.append(d)"
      ],
      "execution_count": 29,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "MSOccuh2NBAB"
      },
      "source": [
        "Resultado da detecção dos keypoints para uma imagem do conjunto das folhas"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JhjA8N_eM8WX",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 286
        },
        "outputId": "05ca989e-fbe6-407e-a3d5-0ed7148310ec"
      },
      "source": [
        "sample = cv2.imread('/content/leafs/leaf1/img1.png')\n",
        "\n",
        "grays = cv2.cvtColor(sample,cv2.COLOR_BGR2GRAY)\n",
        "\n",
        "sift = cv2.xfeatures2d.SIFT_create()\n",
        "\n",
        "keypoints_img1,descriptor_img1 = sift.detectAndCompute(sample,None)\n",
        "\n",
        "sample = cv2.drawKeypoints(sample,keypoints_img1,grays)\n",
        "plt.imshow(sample)"
      ],
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f39b89f66d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 30
        },
        {
          "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": "z81TiV8ANC9J"
      },
      "source": [
        "A próxima etapa consiste na geração das \"visual words\" que serão os bins (posições) do histograma, isto é, representações dos grupos de K \"patches\".\n",
        "\n",
        "No caso específico abaixo, utiliza-se o K-Means para gerar grupos nos vetores de características dos patches. Poderíamos utilizar qualquer outro algoritmo de agrupamento para esse propósito.\n",
        "\n",
        "No caso abaixo, vamos gerar um histograma para a coleção de imagens com 50 \"visual words\". O resultado do K-Means é um modelo representado pelos centróides de cada agrupamento, isto é, as \"visual words\"."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "19jYnG2GNDJB"
      },
      "source": [
        "n_visual_words = 50\n",
        "\n",
        "kmeans_model = KMeans(n_clusters=n_visual_words, \n",
        "                      verbose=False, \n",
        "                      init='random',\n",
        "                      n_init=3)\n",
        "\n",
        "clusters = kmeans_model.fit(dictionary)"
      ],
      "execution_count": 31,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "C0QzGjJhvpRP"
      },
      "source": [
        "Agora temos que gerar o histograma (hist_img) de cada imagem da entrada considerando apenas as \"visual words\" obtidas do K-Means.\n",
        "\n",
        "Para isso, geramos novamente os patches para cada imagem da entrada e contabilizamos a frequência de cada patch no histograma da coleção de imagens (histogram). Observe que cada patch é representado pelo descritor SIFT de 128 posições. O histograma será normalizado em relação ao número de keypoints (pontos locais).\n",
        "\n",
        "Para sabermos em qual bin do histograma que o patch será contabilizado, utilizamos um classificador baseado distância mínima para pegar a label do bin que o patch é mais similar."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "F5kkmYK9NxcH"
      },
      "source": [
        "histogram = []\n",
        "\n",
        "for img_path in l_imgs:\n",
        "\n",
        "  img = cv2.imread(path_imgs+img_path)\n",
        "\n",
        "  keypoints, descriptor = sift.detectAndCompute(img,None)\n",
        "\n",
        "  hist_img = np.zeros(n_visual_words)\n",
        "  number_keypoints = np.size(keypoints)\n",
        "\n",
        "  for d in descriptor:\n",
        "    label = clusters.predict([d])\n",
        "    hist_img[label] += 1/number_keypoints\n",
        "  \n",
        "  histogram.append(hist_img)"
      ],
      "execution_count": 32,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w4XYp1QjxuOK"
      },
      "source": [
        "Por fim, geramos o DataFrame contendo os vetores de características da coleção de imagens da entrada"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "EfBwlGPKQ_Ax"
      },
      "source": [
        "X = np.array(histogram)\n",
        "\n",
        "attributes = []\n",
        "\n",
        "for x in range(1,n_visual_words+1):\n",
        "  attributes.append(\"attrib_\"+str(x))\n",
        "\n",
        "df = pd.DataFrame(X,columns=attributes)"
      ],
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0bGUgzAOzq9K"
      },
      "source": [
        "Resultado final:"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5eLEQDGyzskJ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 437
        },
        "outputId": "c293cb1a-7408-47f2-9806-3891b88ced35"
      },
      "source": [
        "df"
      ],
      "execution_count": 34,
      "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>attrib_1</th>\n",
              "      <th>attrib_2</th>\n",
              "      <th>attrib_3</th>\n",
              "      <th>attrib_4</th>\n",
              "      <th>attrib_5</th>\n",
              "      <th>attrib_6</th>\n",
              "      <th>attrib_7</th>\n",
              "      <th>attrib_8</th>\n",
              "      <th>attrib_9</th>\n",
              "      <th>attrib_10</th>\n",
              "      <th>attrib_11</th>\n",
              "      <th>attrib_12</th>\n",
              "      <th>attrib_13</th>\n",
              "      <th>attrib_14</th>\n",
              "      <th>attrib_15</th>\n",
              "      <th>attrib_16</th>\n",
              "      <th>attrib_17</th>\n",
              "      <th>attrib_18</th>\n",
              "      <th>attrib_19</th>\n",
              "      <th>attrib_20</th>\n",
              "      <th>attrib_21</th>\n",
              "      <th>attrib_22</th>\n",
              "      <th>attrib_23</th>\n",
              "      <th>attrib_24</th>\n",
              "      <th>attrib_25</th>\n",
              "      <th>attrib_26</th>\n",
              "      <th>attrib_27</th>\n",
              "      <th>attrib_28</th>\n",
              "      <th>attrib_29</th>\n",
              "      <th>attrib_30</th>\n",
              "      <th>attrib_31</th>\n",
              "      <th>attrib_32</th>\n",
              "      <th>attrib_33</th>\n",
              "      <th>attrib_34</th>\n",
              "      <th>attrib_35</th>\n",
              "      <th>attrib_36</th>\n",
              "      <th>attrib_37</th>\n",
              "      <th>attrib_38</th>\n",
              "      <th>attrib_39</th>\n",
              "      <th>attrib_40</th>\n",
              "      <th>attrib_41</th>\n",
              "      <th>attrib_42</th>\n",
              "      <th>attrib_43</th>\n",
              "      <th>attrib_44</th>\n",
              "      <th>attrib_45</th>\n",
              "      <th>attrib_46</th>\n",
              "      <th>attrib_47</th>\n",
              "      <th>attrib_48</th>\n",
              "      <th>attrib_49</th>\n",
              "      <th>attrib_50</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>0.009189</td>\n",
              "      <td>0.032029</td>\n",
              "      <td>0.014965</td>\n",
              "      <td>0.033342</td>\n",
              "      <td>0.024678</td>\n",
              "      <td>0.009451</td>\n",
              "      <td>0.014702</td>\n",
              "      <td>0.065897</td>\n",
              "      <td>0.011814</td>\n",
              "      <td>0.015752</td>\n",
              "      <td>0.016015</td>\n",
              "      <td>0.008139</td>\n",
              "      <td>0.008401</td>\n",
              "      <td>0.012602</td>\n",
              "      <td>0.024678</td>\n",
              "      <td>0.016540</td>\n",
              "      <td>0.017065</td>\n",
              "      <td>0.013652</td>\n",
              "      <td>0.032817</td>\n",
              "      <td>0.032292</td>\n",
              "      <td>0.006301</td>\n",
              "      <td>0.018903</td>\n",
              "      <td>0.008664</td>\n",
              "      <td>0.027041</td>\n",
              "      <td>0.012077</td>\n",
              "      <td>0.026254</td>\n",
              "      <td>0.037805</td>\n",
              "      <td>0.018640</td>\n",
              "      <td>0.019428</td>\n",
              "      <td>0.026779</td>\n",
              "      <td>0.007088</td>\n",
              "      <td>0.008401</td>\n",
              "      <td>0.013127</td>\n",
              "      <td>0.016277</td>\n",
              "      <td>0.022578</td>\n",
              "      <td>0.015752</td>\n",
              "      <td>0.008139</td>\n",
              "      <td>0.016277</td>\n",
              "      <td>0.018378</td>\n",
              "      <td>0.019428</td>\n",
              "      <td>0.015490</td>\n",
              "      <td>0.018378</td>\n",
              "      <td>0.012864</td>\n",
              "      <td>0.044894</td>\n",
              "      <td>0.035180</td>\n",
              "      <td>0.014439</td>\n",
              "      <td>0.007876</td>\n",
              "      <td>0.023891</td>\n",
              "      <td>0.055133</td>\n",
              "      <td>0.010501</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>0.007027</td>\n",
              "      <td>0.034024</td>\n",
              "      <td>0.024408</td>\n",
              "      <td>0.041050</td>\n",
              "      <td>0.036982</td>\n",
              "      <td>0.009246</td>\n",
              "      <td>0.006287</td>\n",
              "      <td>0.066568</td>\n",
              "      <td>0.012944</td>\n",
              "      <td>0.018121</td>\n",
              "      <td>0.013314</td>\n",
              "      <td>0.008506</td>\n",
              "      <td>0.005917</td>\n",
              "      <td>0.012944</td>\n",
              "      <td>0.026627</td>\n",
              "      <td>0.012574</td>\n",
              "      <td>0.019970</td>\n",
              "      <td>0.016272</td>\n",
              "      <td>0.023669</td>\n",
              "      <td>0.038831</td>\n",
              "      <td>0.007396</td>\n",
              "      <td>0.029956</td>\n",
              "      <td>0.005917</td>\n",
              "      <td>0.036982</td>\n",
              "      <td>0.012574</td>\n",
              "      <td>0.039201</td>\n",
              "      <td>0.032175</td>\n",
              "      <td>0.012574</td>\n",
              "      <td>0.024038</td>\n",
              "      <td>0.018491</td>\n",
              "      <td>0.007027</td>\n",
              "      <td>0.009615</td>\n",
              "      <td>0.011834</td>\n",
              "      <td>0.011095</td>\n",
              "      <td>0.012574</td>\n",
              "      <td>0.022929</td>\n",
              "      <td>0.007396</td>\n",
              "      <td>0.013314</td>\n",
              "      <td>0.013314</td>\n",
              "      <td>0.015902</td>\n",
              "      <td>0.015902</td>\n",
              "      <td>0.011095</td>\n",
              "      <td>0.013683</td>\n",
              "      <td>0.033284</td>\n",
              "      <td>0.041790</td>\n",
              "      <td>0.021820</td>\n",
              "      <td>0.008876</td>\n",
              "      <td>0.019601</td>\n",
              "      <td>0.045118</td>\n",
              "      <td>0.009246</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>0.005714</td>\n",
              "      <td>0.023673</td>\n",
              "      <td>0.018776</td>\n",
              "      <td>0.045714</td>\n",
              "      <td>0.050612</td>\n",
              "      <td>0.007347</td>\n",
              "      <td>0.008163</td>\n",
              "      <td>0.067755</td>\n",
              "      <td>0.021224</td>\n",
              "      <td>0.038367</td>\n",
              "      <td>0.011429</td>\n",
              "      <td>0.017959</td>\n",
              "      <td>0.021224</td>\n",
              "      <td>0.012245</td>\n",
              "      <td>0.013061</td>\n",
              "      <td>0.014694</td>\n",
              "      <td>0.020408</td>\n",
              "      <td>0.017959</td>\n",
              "      <td>0.006531</td>\n",
              "      <td>0.031020</td>\n",
              "      <td>0.016327</td>\n",
              "      <td>0.017143</td>\n",
              "      <td>0.004898</td>\n",
              "      <td>0.053878</td>\n",
              "      <td>0.017959</td>\n",
              "      <td>0.052245</td>\n",
              "      <td>0.011429</td>\n",
              "      <td>0.004898</td>\n",
              "      <td>0.007347</td>\n",
              "      <td>0.000816</td>\n",
              "      <td>0.014694</td>\n",
              "      <td>0.003265</td>\n",
              "      <td>0.026122</td>\n",
              "      <td>0.007347</td>\n",
              "      <td>0.004082</td>\n",
              "      <td>0.021224</td>\n",
              "      <td>0.008980</td>\n",
              "      <td>0.006531</td>\n",
              "      <td>0.007347</td>\n",
              "      <td>0.013878</td>\n",
              "      <td>0.028571</td>\n",
              "      <td>0.022041</td>\n",
              "      <td>0.013878</td>\n",
              "      <td>0.013061</td>\n",
              "      <td>0.058776</td>\n",
              "      <td>0.026122</td>\n",
              "      <td>0.009796</td>\n",
              "      <td>0.021224</td>\n",
              "      <td>0.039184</td>\n",
              "      <td>0.013061</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>0.013453</td>\n",
              "      <td>0.025112</td>\n",
              "      <td>0.021749</td>\n",
              "      <td>0.034081</td>\n",
              "      <td>0.025785</td>\n",
              "      <td>0.014574</td>\n",
              "      <td>0.026682</td>\n",
              "      <td>0.045740</td>\n",
              "      <td>0.013453</td>\n",
              "      <td>0.018161</td>\n",
              "      <td>0.020404</td>\n",
              "      <td>0.010314</td>\n",
              "      <td>0.006502</td>\n",
              "      <td>0.017713</td>\n",
              "      <td>0.024215</td>\n",
              "      <td>0.016592</td>\n",
              "      <td>0.013901</td>\n",
              "      <td>0.020404</td>\n",
              "      <td>0.010314</td>\n",
              "      <td>0.033632</td>\n",
              "      <td>0.013004</td>\n",
              "      <td>0.030045</td>\n",
              "      <td>0.019507</td>\n",
              "      <td>0.025785</td>\n",
              "      <td>0.009417</td>\n",
              "      <td>0.029372</td>\n",
              "      <td>0.026906</td>\n",
              "      <td>0.016143</td>\n",
              "      <td>0.027130</td>\n",
              "      <td>0.017713</td>\n",
              "      <td>0.011883</td>\n",
              "      <td>0.011659</td>\n",
              "      <td>0.013677</td>\n",
              "      <td>0.023543</td>\n",
              "      <td>0.024664</td>\n",
              "      <td>0.021076</td>\n",
              "      <td>0.008969</td>\n",
              "      <td>0.015247</td>\n",
              "      <td>0.018161</td>\n",
              "      <td>0.022646</td>\n",
              "      <td>0.018834</td>\n",
              "      <td>0.016143</td>\n",
              "      <td>0.013229</td>\n",
              "      <td>0.024215</td>\n",
              "      <td>0.035426</td>\n",
              "      <td>0.011883</td>\n",
              "      <td>0.008744</td>\n",
              "      <td>0.028924</td>\n",
              "      <td>0.032287</td>\n",
              "      <td>0.010987</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>0.007220</td>\n",
              "      <td>0.018051</td>\n",
              "      <td>0.010229</td>\n",
              "      <td>0.032491</td>\n",
              "      <td>0.093863</td>\n",
              "      <td>0.003610</td>\n",
              "      <td>0.012034</td>\n",
              "      <td>0.067389</td>\n",
              "      <td>0.023466</td>\n",
              "      <td>0.011432</td>\n",
              "      <td>0.008424</td>\n",
              "      <td>0.013237</td>\n",
              "      <td>0.010830</td>\n",
              "      <td>0.015644</td>\n",
              "      <td>0.010830</td>\n",
              "      <td>0.012635</td>\n",
              "      <td>0.015644</td>\n",
              "      <td>0.010830</td>\n",
              "      <td>0.043321</td>\n",
              "      <td>0.018652</td>\n",
              "      <td>0.003008</td>\n",
              "      <td>0.014440</td>\n",
              "      <td>0.003008</td>\n",
              "      <td>0.043321</td>\n",
              "      <td>0.012635</td>\n",
              "      <td>0.083634</td>\n",
              "      <td>0.022864</td>\n",
              "      <td>0.007822</td>\n",
              "      <td>0.004813</td>\n",
              "      <td>0.019856</td>\n",
              "      <td>0.012635</td>\n",
              "      <td>0.007822</td>\n",
              "      <td>0.014440</td>\n",
              "      <td>0.011432</td>\n",
              "      <td>0.015042</td>\n",
              "      <td>0.032491</td>\n",
              "      <td>0.005415</td>\n",
              "      <td>0.015644</td>\n",
              "      <td>0.002407</td>\n",
              "      <td>0.011432</td>\n",
              "      <td>0.020457</td>\n",
              "      <td>0.016847</td>\n",
              "      <td>0.009627</td>\n",
              "      <td>0.028279</td>\n",
              "      <td>0.035499</td>\n",
              "      <td>0.020457</td>\n",
              "      <td>0.009627</td>\n",
              "      <td>0.015042</td>\n",
              "      <td>0.054152</td>\n",
              "      <td>0.006017</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>70</th>\n",
              "      <td>0.011583</td>\n",
              "      <td>0.032336</td>\n",
              "      <td>0.011583</td>\n",
              "      <td>0.045367</td>\n",
              "      <td>0.043919</td>\n",
              "      <td>0.005792</td>\n",
              "      <td>0.016409</td>\n",
              "      <td>0.102799</td>\n",
              "      <td>0.016409</td>\n",
              "      <td>0.023649</td>\n",
              "      <td>0.010618</td>\n",
              "      <td>0.011583</td>\n",
              "      <td>0.012066</td>\n",
              "      <td>0.013514</td>\n",
              "      <td>0.014961</td>\n",
              "      <td>0.014961</td>\n",
              "      <td>0.014479</td>\n",
              "      <td>0.007722</td>\n",
              "      <td>0.018822</td>\n",
              "      <td>0.036680</td>\n",
              "      <td>0.010135</td>\n",
              "      <td>0.020270</td>\n",
              "      <td>0.003861</td>\n",
              "      <td>0.037645</td>\n",
              "      <td>0.016409</td>\n",
              "      <td>0.047297</td>\n",
              "      <td>0.018340</td>\n",
              "      <td>0.006757</td>\n",
              "      <td>0.009170</td>\n",
              "      <td>0.009170</td>\n",
              "      <td>0.013031</td>\n",
              "      <td>0.003378</td>\n",
              "      <td>0.018340</td>\n",
              "      <td>0.011100</td>\n",
              "      <td>0.008687</td>\n",
              "      <td>0.016409</td>\n",
              "      <td>0.009653</td>\n",
              "      <td>0.009170</td>\n",
              "      <td>0.010618</td>\n",
              "      <td>0.012548</td>\n",
              "      <td>0.018340</td>\n",
              "      <td>0.020270</td>\n",
              "      <td>0.015927</td>\n",
              "      <td>0.019788</td>\n",
              "      <td>0.037162</td>\n",
              "      <td>0.022683</td>\n",
              "      <td>0.013996</td>\n",
              "      <td>0.015444</td>\n",
              "      <td>0.069015</td>\n",
              "      <td>0.010135</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>71</th>\n",
              "      <td>0.009026</td>\n",
              "      <td>0.031593</td>\n",
              "      <td>0.005158</td>\n",
              "      <td>0.040619</td>\n",
              "      <td>0.069632</td>\n",
              "      <td>0.003224</td>\n",
              "      <td>0.010961</td>\n",
              "      <td>0.104449</td>\n",
              "      <td>0.010316</td>\n",
              "      <td>0.018698</td>\n",
              "      <td>0.011605</td>\n",
              "      <td>0.012250</td>\n",
              "      <td>0.013540</td>\n",
              "      <td>0.011605</td>\n",
              "      <td>0.014184</td>\n",
              "      <td>0.011605</td>\n",
              "      <td>0.013540</td>\n",
              "      <td>0.003868</td>\n",
              "      <td>0.000645</td>\n",
              "      <td>0.031593</td>\n",
              "      <td>0.008382</td>\n",
              "      <td>0.013540</td>\n",
              "      <td>0.003868</td>\n",
              "      <td>0.032237</td>\n",
              "      <td>0.019987</td>\n",
              "      <td>0.090909</td>\n",
              "      <td>0.020632</td>\n",
              "      <td>0.005803</td>\n",
              "      <td>0.003224</td>\n",
              "      <td>0.005803</td>\n",
              "      <td>0.010961</td>\n",
              "      <td>0.005158</td>\n",
              "      <td>0.018053</td>\n",
              "      <td>0.010961</td>\n",
              "      <td>0.014829</td>\n",
              "      <td>0.036750</td>\n",
              "      <td>0.007092</td>\n",
              "      <td>0.003868</td>\n",
              "      <td>0.000645</td>\n",
              "      <td>0.009671</td>\n",
              "      <td>0.025790</td>\n",
              "      <td>0.019342</td>\n",
              "      <td>0.012895</td>\n",
              "      <td>0.037395</td>\n",
              "      <td>0.030948</td>\n",
              "      <td>0.017408</td>\n",
              "      <td>0.012250</td>\n",
              "      <td>0.018698</td>\n",
              "      <td>0.068988</td>\n",
              "      <td>0.005803</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>72</th>\n",
              "      <td>0.011089</td>\n",
              "      <td>0.039604</td>\n",
              "      <td>0.009901</td>\n",
              "      <td>0.047129</td>\n",
              "      <td>0.037624</td>\n",
              "      <td>0.005545</td>\n",
              "      <td>0.010693</td>\n",
              "      <td>0.060594</td>\n",
              "      <td>0.010297</td>\n",
              "      <td>0.019010</td>\n",
              "      <td>0.007525</td>\n",
              "      <td>0.012673</td>\n",
              "      <td>0.009505</td>\n",
              "      <td>0.013069</td>\n",
              "      <td>0.032871</td>\n",
              "      <td>0.010297</td>\n",
              "      <td>0.017030</td>\n",
              "      <td>0.018218</td>\n",
              "      <td>0.011485</td>\n",
              "      <td>0.032871</td>\n",
              "      <td>0.009505</td>\n",
              "      <td>0.025347</td>\n",
              "      <td>0.015050</td>\n",
              "      <td>0.036832</td>\n",
              "      <td>0.013069</td>\n",
              "      <td>0.034059</td>\n",
              "      <td>0.020198</td>\n",
              "      <td>0.009505</td>\n",
              "      <td>0.013069</td>\n",
              "      <td>0.013465</td>\n",
              "      <td>0.012277</td>\n",
              "      <td>0.007525</td>\n",
              "      <td>0.015842</td>\n",
              "      <td>0.014257</td>\n",
              "      <td>0.033267</td>\n",
              "      <td>0.029307</td>\n",
              "      <td>0.008317</td>\n",
              "      <td>0.012673</td>\n",
              "      <td>0.019010</td>\n",
              "      <td>0.015050</td>\n",
              "      <td>0.040792</td>\n",
              "      <td>0.023762</td>\n",
              "      <td>0.010297</td>\n",
              "      <td>0.028515</td>\n",
              "      <td>0.045545</td>\n",
              "      <td>0.015446</td>\n",
              "      <td>0.010693</td>\n",
              "      <td>0.015842</td>\n",
              "      <td>0.026535</td>\n",
              "      <td>0.007921</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>73</th>\n",
              "      <td>0.015718</td>\n",
              "      <td>0.028789</td>\n",
              "      <td>0.030940</td>\n",
              "      <td>0.040371</td>\n",
              "      <td>0.013898</td>\n",
              "      <td>0.014560</td>\n",
              "      <td>0.021013</td>\n",
              "      <td>0.049305</td>\n",
              "      <td>0.009927</td>\n",
              "      <td>0.011747</td>\n",
              "      <td>0.023991</td>\n",
              "      <td>0.009265</td>\n",
              "      <td>0.005956</td>\n",
              "      <td>0.011747</td>\n",
              "      <td>0.015387</td>\n",
              "      <td>0.011913</td>\n",
              "      <td>0.011747</td>\n",
              "      <td>0.013567</td>\n",
              "      <td>0.023825</td>\n",
              "      <td>0.039543</td>\n",
              "      <td>0.006618</td>\n",
              "      <td>0.021840</td>\n",
              "      <td>0.020847</td>\n",
              "      <td>0.021509</td>\n",
              "      <td>0.007445</td>\n",
              "      <td>0.019523</td>\n",
              "      <td>0.040702</td>\n",
              "      <td>0.019358</td>\n",
              "      <td>0.036400</td>\n",
              "      <td>0.039709</td>\n",
              "      <td>0.007611</td>\n",
              "      <td>0.010920</td>\n",
              "      <td>0.010920</td>\n",
              "      <td>0.023494</td>\n",
              "      <td>0.024487</td>\n",
              "      <td>0.018531</td>\n",
              "      <td>0.006453</td>\n",
              "      <td>0.030609</td>\n",
              "      <td>0.025811</td>\n",
              "      <td>0.026473</td>\n",
              "      <td>0.013567</td>\n",
              "      <td>0.011251</td>\n",
              "      <td>0.010093</td>\n",
              "      <td>0.037889</td>\n",
              "      <td>0.022998</td>\n",
              "      <td>0.011085</td>\n",
              "      <td>0.005625</td>\n",
              "      <td>0.019854</td>\n",
              "      <td>0.038882</td>\n",
              "      <td>0.006287</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>74</th>\n",
              "      <td>0.003626</td>\n",
              "      <td>0.012328</td>\n",
              "      <td>0.011603</td>\n",
              "      <td>0.034083</td>\n",
              "      <td>0.095722</td>\n",
              "      <td>0.006526</td>\n",
              "      <td>0.015228</td>\n",
              "      <td>0.059463</td>\n",
              "      <td>0.013053</td>\n",
              "      <td>0.029007</td>\n",
              "      <td>0.010877</td>\n",
              "      <td>0.010152</td>\n",
              "      <td>0.014503</td>\n",
              "      <td>0.022480</td>\n",
              "      <td>0.021030</td>\n",
              "      <td>0.012328</td>\n",
              "      <td>0.023205</td>\n",
              "      <td>0.005076</td>\n",
              "      <td>0.004351</td>\n",
              "      <td>0.019579</td>\n",
              "      <td>0.010152</td>\n",
              "      <td>0.056563</td>\n",
              "      <td>0.002901</td>\n",
              "      <td>0.056563</td>\n",
              "      <td>0.018129</td>\n",
              "      <td>0.069616</td>\n",
              "      <td>0.022480</td>\n",
              "      <td>0.005801</td>\n",
              "      <td>0.002175</td>\n",
              "      <td>0.002901</td>\n",
              "      <td>0.023930</td>\n",
              "      <td>0.006526</td>\n",
              "      <td>0.018129</td>\n",
              "      <td>0.012328</td>\n",
              "      <td>0.002175</td>\n",
              "      <td>0.031182</td>\n",
              "      <td>0.003626</td>\n",
              "      <td>0.002175</td>\n",
              "      <td>0.001450</td>\n",
              "      <td>0.008702</td>\n",
              "      <td>0.005801</td>\n",
              "      <td>0.023930</td>\n",
              "      <td>0.009427</td>\n",
              "      <td>0.019579</td>\n",
              "      <td>0.037708</td>\n",
              "      <td>0.017404</td>\n",
              "      <td>0.012328</td>\n",
              "      <td>0.025381</td>\n",
              "      <td>0.061639</td>\n",
              "      <td>0.005076</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>75 rows × 50 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "    attrib_1  attrib_2  attrib_3  ...  attrib_48  attrib_49  attrib_50\n",
              "0   0.009189  0.032029  0.014965  ...   0.023891   0.055133   0.010501\n",
              "1   0.007027  0.034024  0.024408  ...   0.019601   0.045118   0.009246\n",
              "2   0.005714  0.023673  0.018776  ...   0.021224   0.039184   0.013061\n",
              "3   0.013453  0.025112  0.021749  ...   0.028924   0.032287   0.010987\n",
              "4   0.007220  0.018051  0.010229  ...   0.015042   0.054152   0.006017\n",
              "..       ...       ...       ...  ...        ...        ...        ...\n",
              "70  0.011583  0.032336  0.011583  ...   0.015444   0.069015   0.010135\n",
              "71  0.009026  0.031593  0.005158  ...   0.018698   0.068988   0.005803\n",
              "72  0.011089  0.039604  0.009901  ...   0.015842   0.026535   0.007921\n",
              "73  0.015718  0.028789  0.030940  ...   0.019854   0.038882   0.006287\n",
              "74  0.003626  0.012328  0.011603  ...   0.025381   0.061639   0.005076\n",
              "\n",
              "[75 rows x 50 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 34
        }
      ]
    }
  ]
}