t-snes on Fashion MNSIT.ipynb

{
  "cells": [
    {
      "metadata": {
        "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
        "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
        "trusted": true,
        "scrolled": true
      },
      "cell_type": "code",
      "source": "# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load in \n\n# Input data files are available in the \"../input/\" directory.\n# For example, running this (by clicking run or pressing Shift+Enter) will list the files in the input directory\n\nimport pandas as pd\nimport numpy as np\n\nimport os\nimport tensorflow as tf\nfrom sklearn.manifold import TSNE\nfrom subprocess import check_output\n\n\n# Any results you write to the current directory are saved as output.",
      "execution_count": 1,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "scrolled": true,
        "_uuid": "38243e5ace4bd2601b52f630335ece816d88ce38"
      },
      "cell_type": "code",
      "source": "print(tf.__version__)\n!python -V\n",
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": "1.11.0-rc1\nPython 3.6.6 :: Anaconda, Inc.\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "scrolled": true,
        "_uuid": "eeff057c284cd6a52d49f09214941d9a13df5508"
      },
      "cell_type": "code",
      "source": "#print(os.listdir(\"../input\"))\nprint(check_output([\"ls\", \"../input\"]).decode(\"utf8\"))\n",
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": "fashion-mnist_test.csv\nfashion-mnist_train.csv\nt10k-images-idx3-ubyte\nt10k-labels-idx1-ubyte\ntrain-images-idx3-ubyte\ntrain-labels-idx1-ubyte\n\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "ff4892c05c1fe40336046c47d1c2ecdee4fb55e4"
      },
      "cell_type": "code",
      "source": "data_train_file = \"../input/fashion-mnist_train.csv\"\ndata_test_file = \"../input/fashion-mnist_test.csv\"\n\ndf_train = pd.read_csv(data_train_file)\ndf_test = pd.read_csv(data_test_file)\n",
      "execution_count": 5,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "1b71c58e2fab8ad7126f101339ddb1a2488ac704"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "5bd10d2d665f0acfbb2c293b9a30485bf0f67649"
      },
      "cell_type": "code",
      "source": "df_train.head()",
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 6,
          "data": {
            "text/plain": "   label  pixel1  pixel2    ...     pixel782  pixel783  pixel784\n0      2       0       0    ...            0         0         0\n1      9       0       0    ...            0         0         0\n2      6       0       0    ...            0         0         0\n3      0       0       0    ...            0         0         0\n4      3       0       0    ...            0         0         0\n\n[5 rows x 785 columns]",
            "text/html": "<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>label</th>\n      <th>pixel1</th>\n      <th>pixel2</th>\n      <th>pixel3</th>\n      <th>pixel4</th>\n      <th>pixel5</th>\n      <th>pixel6</th>\n      <th>pixel7</th>\n      <th>pixel8</th>\n      <th>pixel9</th>\n      <th>pixel10</th>\n      <th>pixel11</th>\n      <th>pixel12</th>\n      <th>pixel13</th>\n      <th>pixel14</th>\n      <th>pixel15</th>\n      <th>pixel16</th>\n      <th>pixel17</th>\n      <th>pixel18</th>\n      <th>pixel19</th>\n      <th>pixel20</th>\n      <th>pixel21</th>\n      <th>pixel22</th>\n      <th>pixel23</th>\n      <th>pixel24</th>\n      <th>pixel25</th>\n      <th>pixel26</th>\n      <th>pixel27</th>\n      <th>pixel28</th>\n      <th>pixel29</th>\n      <th>pixel30</th>\n      <th>pixel31</th>\n      <th>pixel32</th>\n      <th>pixel33</th>\n      <th>pixel34</th>\n      <th>pixel35</th>\n      <th>pixel36</th>\n      <th>pixel37</th>\n      <th>pixel38</th>\n      <th>pixel39</th>\n      <th>...</th>\n      <th>pixel745</th>\n      <th>pixel746</th>\n      <th>pixel747</th>\n      <th>pixel748</th>\n      <th>pixel749</th>\n      <th>pixel750</th>\n      <th>pixel751</th>\n      <th>pixel752</th>\n      <th>pixel753</th>\n      <th>pixel754</th>\n      <th>pixel755</th>\n      <th>pixel756</th>\n      <th>pixel757</th>\n      <th>pixel758</th>\n      <th>pixel759</th>\n      <th>pixel760</th>\n      <th>pixel761</th>\n      <th>pixel762</th>\n      <th>pixel763</th>\n      <th>pixel764</th>\n      <th>pixel765</th>\n      <th>pixel766</th>\n      <th>pixel767</th>\n      <th>pixel768</th>\n      <th>pixel769</th>\n      <th>pixel770</th>\n      <th>pixel771</th>\n      <th>pixel772</th>\n      <th>pixel773</th>\n      <th>pixel774</th>\n      <th>pixel775</th>\n      <th>pixel776</th>\n      <th>pixel777</th>\n      <th>pixel778</th>\n      <th>pixel779</th>\n      <th>pixel780</th>\n      <th>pixel781</th>\n      <th>pixel782</th>\n      <th>pixel783</th>\n      <th>pixel784</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>9</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>...</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>6</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>5</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>105</td>\n      <td>92</td>\n      <td>101</td>\n      <td>107</td>\n      <td>100</td>\n      <td>132</td>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n      <td>4</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>150</td>\n      <td>...</td>\n      <td>211</td>\n      <td>220</td>\n      <td>214</td>\n      <td>74</td>\n      <td>0</td>\n      <td>255</td>\n      <td>222</td>\n      <td>128</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>44</td>\n      <td>12</td>\n      <td>0</td>\n      <td>0</td>\n      <td>40</td>\n      <td>134</td>\n      <td>162</td>\n      <td>191</td>\n      <td>214</td>\n      <td>163</td>\n      <td>146</td>\n      <td>165</td>\n      <td>79</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>30</td>\n      <td>43</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>2</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>114</td>\n      <td>183</td>\n      <td>112</td>\n      <td>55</td>\n      <td>23</td>\n      <td>72</td>\n      <td>102</td>\n      <td>165</td>\n      <td>160</td>\n      <td>28</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>24</td>\n      <td>188</td>\n      <td>163</td>\n      <td>93</td>\n      <td>...</td>\n      <td>171</td>\n      <td>249</td>\n      <td>207</td>\n      <td>197</td>\n      <td>202</td>\n      <td>45</td>\n      <td>0</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>22</td>\n      <td>21</td>\n      <td>25</td>\n      <td>69</td>\n      <td>52</td>\n      <td>45</td>\n      <td>74</td>\n      <td>39</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>46</td>\n      <td>0</td>\n      <td>21</td>\n      <td>68</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>25</td>\n      <td>187</td>\n      <td>189</td>\n      <td>...</td>\n      <td>230</td>\n      <td>237</td>\n      <td>229</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>68</td>\n      <td>116</td>\n      <td>112</td>\n      <td>136</td>\n      <td>147</td>\n      <td>144</td>\n      <td>121</td>\n      <td>102</td>\n      <td>63</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
        "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a",
        "trusted": true,
        "scrolled": false
      },
      "cell_type": "code",
      "source": "# Select all columns but the first\nfeatures = df_train[df_train.columns[1:]]\nprint(features.head())\nprint(features.values.shape)",
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": "   pixel1  pixel2  pixel3    ...     pixel782  pixel783  pixel784\n0       0       0       0    ...            0         0         0\n1       0       0       0    ...            0         0         0\n2       0       0       0    ...            0         0         0\n3       0       0       0    ...            0         0         0\n4       0       0       0    ...            0         0         0\n\n[5 rows x 784 columns]\n(60000, 784)\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "9ab2e77ce6f01d13639e2e101c31676f7204ea12"
      },
      "cell_type": "code",
      "source": "labels = df_train['label']\nlabels.describe(include='all')",
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 9,
          "data": {
            "text/plain": "count    60000.000000\nmean         4.500000\nstd          2.872305\nmin          0.000000\n25%          2.000000\n50%          4.500000\n75%          7.000000\nmax          9.000000\nName: label, dtype: float64"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "5b32ca8dbe5346e9b56d5259d0cbf75ceb965060"
      },
      "cell_type": "code",
      "source": "features.values.shape",
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 13,
          "data": {
            "text/plain": "(60000, 784)"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "98ad614f85581a58c8773d0033ed4fc198f0d1a1"
      },
      "cell_type": "code",
      "source": "train_input_fn = tf.estimator.inputs.numpy_input_fn(\n        x={\"pixels\": features.values/255},\n        y=labels,\n        batch_size=100,\n        num_epochs=3,\n        shuffle=True)",
      "execution_count": 19,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "92fb34020d1a3d231d32b66ce22206784581d91a"
      },
      "cell_type": "code",
      "source": "feature_columns = [tf.feature_column.numeric_column(\"pixels\", shape=784)]",
      "execution_count": 20,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "8ac81ef700896338e44ec0a50140fa87e87febd2"
      },
      "cell_type": "code",
      "source": "classifier = tf.estimator.LinearClassifier(\n                feature_columns=feature_columns, \n                n_classes=10,\n                model_dir=\"./models/linear1\"\n                )",
      "execution_count": 21,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "c32d465b5908ec7238446d537dadeac20b72db37"
      },
      "cell_type": "code",
      "source": "classifier.train(input_fn=train_input_fn)\n",
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 22,
          "data": {
            "text/plain": "<tensorflow.python.estimator.canned.linear.LinearClassifier at 0x7f52372dd668>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "b65ce299baa18a7fec3f4022a37c2c93883adf6b"
      },
      "cell_type": "code",
      "source": "features = df_test[df_test.columns[1:]]\nlabels = df_test[\"label\"]\n\nevaluate_input_fn = tf.estimator.inputs.numpy_input_fn(\n        x={\"pixels\": features.values/255},\n        y=labels,\n        batch_size=100,\n        num_epochs=1,\n        shuffle=False)",
      "execution_count": 23,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "79266ee680493b9afa10e0b50992743c37ee362e"
      },
      "cell_type": "code",
      "source": "classifier.evaluate(input_fn=evaluate_input_fn)[\"accuracy\"]",
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 24,
          "data": {
            "text/plain": "0.8398"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "595eb443c72a7d23ff9f5c96396ef189d0bbd8b2"
      },
      "cell_type": "code",
      "source": "# Bonus round 1: predictions\nfeatures = df_test[df_test.columns[1:]]\nlabels = df_test[\"label\"]\n\npredict_input_fn = tf.estimator.inputs.numpy_input_fn(        \n        x={'pixels': features.iloc[5000:5005].values/255},\n        batch_size=1,\n        num_epochs=1,\n        shuffle=False)\npredictions = classifier.predict(input_fn=predict_input_fn)\n\nfor prediction in predictions:\n    print(\"Predictions:    {} with probabilities {}\\n\".format(\n        prediction[\"classes\"], prediction[\"probabilities\"]))\nprint('Expected answers values: \\n{}'.format(\n    labels.iloc[5000:5005]))",
      "execution_count": 25,
      "outputs": [
        {
          "output_type": "stream",
          "text": "Predictions:    [b'1'] with probabilities [5.7030856e-08 9.9999499e-01 4.5030358e-08 2.7750575e-06 2.2156275e-06\n 1.6931392e-21 3.3415102e-09 5.3895652e-18 2.6036551e-10 1.8301781e-18]\n\nPredictions:    [b'6'] with probabilities [1.8847473e-03 2.6059313e-05 1.8405412e-01 1.0129731e-05 8.5942083e-05\n 3.8973140e-18 8.1388360e-01 7.0660281e-29 5.5438872e-05 6.1384369e-22]\n\nPredictions:    [b'4'] with probabilities [9.5862319e-04 3.5619428e-06 5.7421118e-04 5.6254915e-03 6.3929337e-01\n 1.4365425e-16 3.5325342e-01 2.9444088e-19 2.9138595e-04 1.2339624e-17]\n\nPredictions:    [b'1'] with probabilities [3.8481787e-09 9.9999785e-01 4.3475694e-09 9.1417388e-07 1.2396004e-06\n 2.2802653e-25 6.3667305e-10 7.9862078e-20 2.1958954e-12 1.2811041e-19]\n\nPredictions:    [b'9'] with probabilities [2.0792319e-15 1.7221245e-18 2.2008083e-14 1.2545504e-13 1.8621752e-13\n 2.0534690e-03 6.8902261e-14 5.3718756e-04 4.4737554e-11 9.9740940e-01]\n\nExpected answers values: \n5000    1\n5001    2\n5002    4\n5003    1\n5004    9\nName: label, dtype: int64\n",
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "42b86e603046a3abf7fe5be2044bf07be01cb74b"
      },
      "cell_type": "code",
      "source": "# Bonus round 2: visualizing our predictions\n# Import, setup, and a utility for int->string class conversion\nimport matplotlib.pyplot as plt\n%matplotlib inline\nclass_table = [\n    \"T-shirt/top\",\n    \"Trouser\",\n    \"Pullover\",\n    \"Dress\",\n    \"Coat\",\n    \"Sandal\",\n    \"Shirt\",\n    \"Sneaker\",\n    \"Bag\",\n    \"Ankle boot\"\n]\n\ndef get_label_cls(label):\n    \"\"\"given an int label range [0,9], return the string description of that label\"\"\"\n    return class_table[label]\n\nget_label_cls(3)",
      "execution_count": 26,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 26,
          "data": {
            "text/plain": "'Dress'"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "91afc36db9a605e9068efb011686af7089524562"
      },
      "cell_type": "code",
      "source": "for i in range(5000,5005): \n    sample = np.reshape(df_test[df_test.columns[1:]].iloc[i].values/255, (28,28))\n    plt.figure()\n    plt.title(\"labeled class {}\".format(get_label_cls(df_test[\"label\"].iloc[i])))\n    plt.imshow(sample, 'gray')",
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAEICAYAAACQ6CLfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAFp1JREFUeJzt3XuwnHV9x/H3JyEJud/JHEgwAcK9lkukVhlLq0VkpkWmyKUdikINg9CRlqoUbWWmTksrSulQdaJQQJGLVQpSLgIyUkDA4KQQQBEil8SQcyBALhBy+/aPfWKXw9nf73h29+ye/D6vmZ2zZ7/Pb5/fPsn3PPs8v5siAjMrz6hOV8DMOsPJb1YoJ79ZoZz8ZoVy8psVyslvVignf4tIelbSBwa5bUjaZ4j7Gdayko6StHIo+7Pu5uS3riPpg5LulbReUp+kH0n64xa875WSvtCKOu4MnPzWVSSdAHwHuBqYC8wB/h74o07Wa2fk5G8DSUdI+rGkVyWtlnSZpLH9NjtW0gpJL0n6oqRRdeVPl/SkpFck3SHpHQ32M07SxZKel7RG0tckja+Lf6ra/68knZ6p8wxJ/1Ft+4qk/2qw3fmSnqnOyk9IOr4utk91ln6t+lzXV69L0iWSeiWtk/SYpIMHeG8BXwb+ISK+ERGvRcT2iPhRRHy82maUpM9Jeq56v6slTa17j+9IerGqw72SDqpeXwz8GfBpSRskfT91PIoQEX604AE8C3ygen448G5gF2A+8CRwbt22AdwDzAD2BJ4C/qKKHQc8DRxQlf8c8EC/svtUzy8Bbq7eZzLwfeCfqtgxwBrgYGAi8O36sgPU/7+B64HpwBjg96rXjwJW1m33EWB3aieOk4CNQE8Vuxb4bBXbFTiyev2DwCPANEDVZ+sZoA77V3VckDjOp1fHZy9gEvA94Jv94pOBccC/AsvqYlcCX+j0/5VueXS8AjvLoz75B4idC9xY93sAx9T9/gng7ur5bcAZdbFRwOvAO+rK7lMl0UZg77ptfxf4ZfX8CuCiuti+jZIf6AG2A9MHiL0l+QeILwOOq55fDSwB5vbb5g+o/YF7NzAq8V7vreq4a2Kbu4FP1P2+H7AF2GWAbadV7ze1+t3JX/fw1/42kLSvpFuqr5/rgH8EZvXb7IW6589RO5sCvAO4tLpkeBVYSy3R9+hXfjYwAXikbtvbq9ep3q//PhqZB6yNiFcG8dn+XNKyun0eXPfZPl3V9WFJj++41IiIHwKXAf8O9EpaImnKAG//cvWzJ1GF3ft9lueofUOaI2m0pIuqy5J11P4gw9uPveFr/nb5KvAzYGFETAEuoJYU9ebVPd8T+FX1/AXgzIiYVvcYHxEP9Cv/EvAGcFDddlMjYlIVXz3APhp5AZghaVrqQ1X3Hr4OnAPMjIhpwPIdny0iXoyIj0fE7sCZwFd2NC1GxL9FxOHAgdS+hXxqgF38vKrLnySq8StqfyDrP9dWapc4f0rtsukDwFRql1zw/8feQ1jrOPnbYzKwDtggaX/grAG2+ZSk6ZLmAZ+kdr0N8DXgb+tuVE2V9JH+hSNiO7VEvETSbtW2e0j6YLXJDcBHJR0oaQLw+UaVjYjV1C43vlLVaYyk9w2w6URqCdRX7e9j1M78VL9/RNLc6tdXqm23S3qXpN+RNIbapcomapcZ/esRwF8DfyfpY5KmVDf4jpS0pNrsWuCvJC2QNInat6rrI2IrteP+JrVvEBOqWL011O4VGE7+dvkbameh9dQS9PoBtrmJ2k2wZdRutl0OEBE3Av8MXFd9dV0OfKjBfj5D7ebXg9W2d1G7BiYibqN2w+uH1TY/zNT5VGrXzj8Deqndp3iLiHgC+BLwY2qJ9FvA/XWbvAt4SNIGajciPxkRK4Ap1XF4hdrX9JeBLw5UiYj4T2o3Ek+ndpZfA3yB2vGC2r2MbwL3Ar+k9ofkL6vY1dX7rwKeAB7s9/aXAwdWlywDtmaURNWNEDMrjM/8ZoVy8psVyslvVignv1mhdhnOnUny3cVhNm7cuKbi69atS8ZHjUqfP2bOnNkw1tfXlyxrQxMR/fuUDKip5Jd0DHApMBr4RkRc1Mz77axq41Uaa2eLy/z585PxBQsWJOO33357Mj5+/Phk/KSTTmoYu+yyy5Jlm5U77ikltIIN+Wu/pNHUumt+iFqvrVMkHdiqiplZezVzzX8E8HRErIiIzcB11LpWmtkI0Ezy78FbB46s5O2DT5C0WNJSSUub2JeZtVjbb/hFxBJqwzx9w8+sizRz5l/FW0eNza1eM7MRoJnk/wmwsBpdNRY4mdpgDjMbAYb8tT8itko6B7iDWlPfFRHxeMtqthNpttlozz1TQ/Hh5JNPbhjbe++9k2VPOOGEZPzxx9P/pGPH9p+a8K322qvxCNr99tsvWfaOO+5Ixm+55ZZkvJnj3snm2eHS1DV/RNwK3NqiupjZMHL3XrNCOfnNCuXkNyuUk9+sUE5+s0I5+c0KNawTeI7k7r2pcevbt79tFuq32Gef9KrYixcvTsZ32223ZDy1/w0bNiTLzpqVXs9i4cKFQ943wPPPP98w1tvbmyybGy48e/bsZPyaa65pGLvuuuuSZXO6uR/AYMfz+8xvVignv1mhnPxmhXLymxXKyW9WKCe/WaGGderubpZrusk1aaWcd955yfguu6T/GdauXZuMb926tWEsN+T2scceS8bnzp2bjOea21Lvn/vcuX+TXDPm8ccf3zCW+1wXX3xxMr4zDOn1md+sUE5+s0I5+c0K5eQ3K5ST36xQTn6zQjn5zQrldv4WeM973pOM9/T0JOMrVqxIxnNDW1N9EDZu3JgsO2nSpGT8wQcfTMZff/31ZDzVlr/rrrsmy06cODEZz/W9eOmllxrGDj300GTZEvjMb1YoJ79ZoZz8ZoVy8psVyslvVignv1mhnPxmhXI7fwvsu+++yXiuPXrKlCnJ+GuvvZaMp8bsp6YcBxg9enQynhuvv2nTpmQ8Vfdx48Yly27bti0Z37x5czI+Z86chrHccckti56aknykaCr5JT0LrAe2AVsjYlErKmVm7deKM//vR0TjrlRm1pV8zW9WqGaTP4AfSHpE0oBrTklaLGmppKVN7svMWqjZr/1HRsQqSbsBd0r6WUTcW79BRCwBlsDIXqvPbGfT1Jk/IlZVP3uBG4EjWlEpM2u/ISe/pImSJu94DhwNLG9VxcysvZr52j8HuLGaW30X4NsRcXtLatUBzczDvt9++yXjuTbl3Pz1ubb0VDt/ri091wch19aem1s/99makeujkPpsubkC3M6fEBErgN9uYV3MbBi5qc+sUE5+s0I5+c0K5eQ3K5ST36xQHtLbArvttlsyPnXq1GQ8N6R3/fr1yfirr77aMDZr1qxk2dwS3lu2bEnGc82YqabG3LTiY8aMScZzxz01ZXqumfCwww5Lxu+7775kfCTwmd+sUE5+s0I5+c0K5eQ3K5ST36xQTn6zQjn5zQrldv4WyLU3v/nmm8n4O9/5zmQ81959//33N4zlhgPn2rtzQ4JzbfWpYbW54b65YbcHHXRQMp5afvyZZ55Jlt1///2T8Z2Bz/xmhXLymxXKyW9WKCe/WaGc/GaFcvKbFcrJb1Yot/MP0owZMxrGcu34qXHlANOnT0/GU+P1IT2mPteOn1vmOif3/lu3bm0YSy2hDfDyyy8n47k+CKn37+vrS5adP39+Mr4z8JnfrFBOfrNCOfnNCuXkNyuUk9+sUE5+s0I5+c0K5Xb+QUqN2X/jjTeSZXt7e5Pxww8/PBnPtfO3c4nunNxcA6k1B3J9BHJef/31ZDw1X8Dq1auTZefNmzekOo0k2TO/pCsk9UpaXvfaDEl3SvpF9TPdS8XMus5gvvZfCRzT77XzgbsjYiFwd/W7mY0g2eSPiHuBtf1ePg64qnp+FfDhFtfLzNpsqNf8cyJix0XTi0DDTtSSFgOLh7gfM2uTpm/4RURIikR8CbAEILWdmQ2voTb1rZHUA1D9TN/ONrOuM9Tkvxk4rXp+GnBTa6pjZsMl+7Vf0rXAUcAsSSuBzwMXATdIOgN4DjixnZXsBql53HPzz++6667J+IYNG5Lx3Jj7mTNnNozl5tXPtbVPmDAhGc/NZZDqZ5BbUyA3b3/uuKX6MKTmGRjMe+8MsskfEac0CL2/xXUxs2Hk7r1mhXLymxXKyW9WKCe/WaGc/GaF8pDeQTrggAMaxnJNfXPnzk3Gly1bloznhq6mpv7ODQfOyQ35zX32VFNis1OeP/zww8l4alju7rvvniybawrM/ZuuXLkyGe8GPvObFcrJb1YoJ79ZoZz8ZoVy8psVyslvVignv1mh3M4/SKmpu7dt25Ysm1reG/Lt/LnpsVNt8bm65d57y5YtyXiunT+1fHjuvXPTjufa4lNTqqeGQQM89dRTyXhqiDe4nd/MupiT36xQTn6zQjn5zQrl5DcrlJPfrFBOfrNCuZ1/kFJtzhHphYhybe3r1q1LxptpS8+1lUtKxnOfLSdVt9xcAblpxVPvDel2/ilTpiTL5uZQyE0rPhL4zG9WKCe/WaGc/GaFcvKbFcrJb1YoJ79ZoZz8ZoVyO/8gjR8/vmEs1xaeWyY71R4NMHv27GQ8N649JdcPINfWnmurTy1P/sorryTL5pbJnjVrVjLe19fXMDZ58uSm9p2bo2EkyJ75JV0hqVfS8rrXLpS0StKy6nFse6tpZq02mK/9VwLHDPD6JRFxSPW4tbXVMrN2yyZ/RNwLrB2GupjZMGrmht85kh6tLgsaLhYnabGkpZKWNrEvM2uxoSb/V4G9gUOA1cCXGm0YEUsiYlFELBrivsysDYaU/BGxJiK2RcR24OvAEa2tlpm125CSX1L92snHA8sbbWtm3Snbzi/pWuAoYJaklcDngaMkHQIE8CxwZhvr2BVS47dzbd2bNm1KxqdNm5aM58bcp8b759rxx44dm4zn2vmbmdc/19ae6x+RK586bs30jYD0Og4jRTb5I+KUAV6+vA11MbNh5O69ZoVy8psVyslvVignv1mhnPxmhfKQ3kFKDU3NNcXlmqxyclNU56YGT8lNUZ363IOxefPmhrFcM2Puc+eOe6qJddKkScmyuam9p09v2KN9xPCZ36xQTn6zQjn5zQrl5DcrlJPfrFBOfrNCOfnNCuV2/kpu6Gpq6u5ce/PLL7+cjKfawnP7zpXPtZXn4s1MzQ3pPgi5/gm5ZbBzw3JTU4OvX78+WTbXD8BLdJvZiOXkNyuUk9+sUE5+s0I5+c0K5eQ3K5ST36xQbuevNDuFdcqqVauS8Vx7dWr6a4B169Y1jOXa4XPTiuf6GEyYMCEZf+211xrGckub5/pP5Oq2evXqhrEVK1Yky+amBc/teyTwmd+sUE5+s0I5+c0K5eQ3K5ST36xQTn6zQjn5zQo1mCW65wFXA3OoLcm9JCIulTQDuB6YT22Z7hMjovEA6i6Xaw9PtdXPmjUrWTY3bj03Zj5XPrXEd26ugDFjxiTjuXHruf4PqaWsU+PtIb+mwNSpU4dcPlfvDRs2JOPNLvHdDQZz5t8KnBcRBwLvBs6WdCBwPnB3RCwE7q5+N7MRIpv8EbE6In5aPV8PPAnsARwHXFVtdhXw4XZV0sxa7ze65pc0HzgUeAiYExE7+k++SO2ywMxGiEH37Zc0CfgucG5ErKvvdx0RIWnAjtqSFgOLm62ombXWoM78ksZQS/xrIuJ71ctrJPVU8R6gd6CyEbEkIhZFxKJWVNjMWiOb/Kqd4i8HnoyIL9eFbgZOq56fBtzU+uqZWbsM5mv/e4FTgcckLateuwC4CLhB0hnAc8CJ7ani8MgN0UwNq80t59xuqaGvW7ZsSZZttjkt11yXOq65oco5uSG/qea4VPMopIdJw84xpDd79CPiPqDRUX5/a6tjZsPFPfzMCuXkNyuUk9+sUE5+s0I5+c0K5eQ3K5Sn7q7klmxODW2dOXNmsuyNN96YjJ9++unJeG74aKq9e9y4ccmyGzduTMZz/QByw5FfffXVhrHZs2cny+b6KOT6IKSO2z333JMse8IJJyTjqc81UvjMb1YoJ79ZoZz8ZoVy8psVyslvVignv1mhnPxmhXI7fyXXzv/CCy80jE2fPj1Z9rbbbkvGzzrrrGQ8N/12annx3OfKtbXn5irITXn+9NNPN4zlpg3Pfe7cvlPLg3/rW99Klj311FOT8YceeigZHwl85jcrlJPfrFBOfrNCOfnNCuXkNyuUk9+sUE5+s0K5nX+Qzj777La99xtvvJGM59raU2Pyc+Pxc3Pn5/oJ5Mbcp+qemxs/NxfB2rVrk/HUHAy5sgsWLEjGdwY+85sVyslvVignv1mhnPxmhXLymxXKyW9WKCe/WaGy7fyS5gFXA3OAAJZExKWSLgQ+DvRVm14QEbe2q6I7s0ceeSQZP/roo5PxN998s2Es144/alT673+uj0GuH0EqPmnSpGTZXB+C8ePHJ+MPPPBAMl66wXTy2QqcFxE/lTQZeETSnVXskoi4uH3VM7N2ySZ/RKwGVlfP10t6Etij3RUzs/b6ja75Jc0HDgV2zGF0jqRHJV0hacC5rCQtlrRU0tKmampmLTXo5Jc0CfgucG5ErAO+CuwNHELtm8GXBioXEUsiYlFELGpBfc2sRQaV/JLGUEv8ayLiewARsSYitkXEduDrwBHtq6aZtVo2+VVbAvZy4MmI+HLd6z11mx0PLG999cysXRQR6Q2kI4H/AR4DdqzHfAFwCrWv/AE8C5xZ3RxMvVd6ZyNUaolsgNwxzk2fffPNNyfjqSHB27ZtS5bdtGlTMp6bXjs1bXju/XNlR48enYzPnTs3GV+4cGEyntLsv2knRUS68pXB3O2/DxjozdymbzaCuYefWaGc/GaFcvKbFcrJb1YoJ79ZoZz8ZoXy1N1doK+vLxm/6667kvGenp6Gsdy04Llhs7k+CLm2+NQy27kluHNLn99www3JeDO6uR2/VXzmNyuUk9+sUE5+s0I5+c0K5eQ3K5ST36xQTn6zQmXH87d0Z1If8FzdS7OAl4atAr+Zbq1bt9YLXLehamXd3hER6c4ZlWFN/rftXFrarXP7dWvdurVe4LoNVafq5q/9ZoVy8psVqtPJv6TD+0/p1rp1a73AdRuqjtSto9f8ZtY5nT7zm1mHOPnNCtWR5Jd0jKSfS3pa0vmdqEMjkp6V9JikZZ1eX7BaA7FX0vK612ZIulPSL6qf6UHvw1u3CyWtqo7dMknHdqhu8yTdI+kJSY9L+mT1ekePXaJeHTluw37NL2k08BTwh8BK4CfAKRHxxLBWpAFJzwKLIqLjHUIkvQ/YAFwdEQdXr/0LsDYiLqr+cE6PiM90Sd0uBDZ0etn2ajWpnvpl5YEPAx+lg8cuUa8T6cBx68SZ/wjg6YhYERGbgeuA4zpQj64XEfcCa/u9fBxwVfX8Kmr/eYZdg7p1hYhYHRE/rZ6vB3YsK9/RY5eoV0d0Ivn3AF6o+30lHTwAAwjgB5IekbS405UZwJy6ZdFeBOZ0sjIDyC7bPpz6LSvfNcduKMvdt5pv+L3dkRFxGPAh4Ozq621Xito1Wze11Q5q2fbhMsCy8r/WyWM31OXuW60Tyb8KmFf3+9zqta4QEauqn73AjXTf0uNrdqyQXP3s7XB9fq2blm0faFl5uuDYddNy951I/p8ACyUtkDQWOBlIL0M7TCRNrG7EIGkicDTdt/T4zcBp1fPTgJs6WJe36JZl2xstK0+Hj13XLXcfEcP+AI6ldsf/GeCznahDg3rtBfxv9Xi803UDrqX2NXALtXsjZwAzgbuBXwB3ATO6qG7fpLaU+6PUEq2nQ3U7ktpX+keBZdXj2E4fu0S9OnLc3L3XrFC+4WdWKCe/WaGc/GaFcvKbFcrJb1YoJ79ZoZz8ZoX6P5cQmsPvwKPSAAAAAElFTkSuQmCC\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": "<Figure size 432x288 with 1 Axes>",
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "cfe4de484922e91b753d4f85ae68b3318a25f3f1"
      },
      "cell_type": "code",
      "source": "DNN = tf.estimator.DNNClassifier(\n                feature_columns=feature_columns, \n                hidden_units=[40,30,20],\n                n_classes=10,\n                model_dir=\"./models/deep1\"\n                )\n",
      "execution_count": 29,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "fb6f4ec0f5777f898a598179b443e73b42057b90"
      },
      "cell_type": "code",
      "source": "DNN.train(input_fn=train_input_fn)\n",
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 30,
          "data": {
            "text/plain": "<tensorflow.python.estimator.canned.dnn.DNNClassifier at 0x7f52043b5c18>"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "1ada2501fdf93eef987ae4c8320b9bb9edba2230"
      },
      "cell_type": "code",
      "source": "DNN.evaluate(input_fn=evaluate_input_fn)[\"accuracy\"]",
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 31,
          "data": {
            "text/plain": "0.8058"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "e6a676e773255143dd4182ffa89cc743315340fd"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "712a3d9b26db487d093f51562f6e640e4314435b"
      },
      "cell_type": "code",
      "source": "from sklearn.preprocessing import StandardScaler\nstandardized_data = StandardScaler().fit_transform(df_train)\nstandardized_data.shape",
      "execution_count": 33,
      "outputs": [
        {
          "output_type": "execute_result",
          "execution_count": 33,
          "data": {
            "text/plain": "(60000, 785)"
          },
          "metadata": {}
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "f8dfc9487f400fe1f99428d4e05743757d17e6a2"
      },
      "cell_type": "code",
      "source": "model = TSNE(n_components=2, random_state=0, perplexity=30, learning_rate=200, n_iter=1000)\nreduced_data = model.fit_transform(standardized_data)",
      "execution_count": 34,
      "outputs": [
        {
          "output_type": "error",
          "ename": "KeyboardInterrupt",
          "evalue": "",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-34-e64afd257b5b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mmodel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mTSNE\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mn_components\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m2\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mrandom_state\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mperplexity\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m30\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlearning_rate\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m200\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_iter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m1000\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mreduced_data\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit_transform\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mstandardized_data\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m/opt/conda/lib/python3.6/site-packages/sklearn/manifold/t_sne.py\u001b[0m in \u001b[0;36mfit_transform\u001b[0;34m(self, X, y)\u001b[0m\n\u001b[1;32m    856\u001b[0m             \u001b[0mEmbedding\u001b[0m \u001b[0mof\u001b[0m \u001b[0mthe\u001b[0m \u001b[0mtraining\u001b[0m \u001b[0mdata\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mlow\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0mdimensional\u001b[0m \u001b[0mspace\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    857\u001b[0m         \"\"\"\n\u001b[0;32m--> 858\u001b[0;31m         \u001b[0membedding\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    859\u001b[0m         \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0membedding_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0membedding\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    860\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0membedding_\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/opt/conda/lib/python3.6/site-packages/sklearn/manifold/t_sne.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X, skip_num_points)\u001b[0m\n\u001b[1;32m    715\u001b[0m                                    metric=self.metric)\n\u001b[1;32m    716\u001b[0m             \u001b[0mt0\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 717\u001b[0;31m             \u001b[0mknn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    718\u001b[0m             \u001b[0mduration\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtime\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mt0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    719\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/opt/conda/lib/python3.6/site-packages/sklearn/neighbors/base.py\u001b[0m in \u001b[0;36mfit\u001b[0;34m(self, X, y)\u001b[0m\n\u001b[1;32m    801\u001b[0m             \u001b[0;32mor\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0mn_samples\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mn_samples\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mmetric\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'precomputed'\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    802\u001b[0m         \"\"\"\n\u001b[0;32m--> 803\u001b[0;31m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mX\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;32m/opt/conda/lib/python3.6/site-packages/sklearn/neighbors/base.py\u001b[0m in \u001b[0;36m_fit\u001b[0;34m(self, X)\u001b[0m\n\u001b[1;32m    246\u001b[0m             self._tree = KDTree(X, self.leaf_size,\n\u001b[1;32m    247\u001b[0m                                 \u001b[0mmetric\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0meffective_metric_\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 248\u001b[0;31m                                 **self.effective_metric_params_)\n\u001b[0m\u001b[1;32m    249\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_fit_method\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;34m'brute'\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    250\u001b[0m             \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_tree\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
          ]
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "5cdef5e9be34816949befad04171690045ddf2cb"
      },
      "cell_type": "code",
      "source": "reduced_df = np.vstack((reduced_data.T, label)).T\nreduced_df = pd.DataFrame(data=reduced_df, columns=[\"X\", \"Y\", \"label\"])\nreduced_df.label = reduced_df.label.astype(np.int)\nreduced_df.head()",
      "execution_count": 32,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "name 'reduced_data' is not defined",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-32-e32bd9ff10d6>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mreduced_df\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvstack\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mreduced_data\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mT\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlabel\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mT\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mreduced_df\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mpd\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mDataFrame\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mreduced_df\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mcolumns\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"X\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"Y\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"label\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mreduced_df\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlabel\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mreduced_df\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlabel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mastype\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mint\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mreduced_df\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mhead\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'reduced_data' is not defined"
          ]
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "ff5fbe58605cf11a127ee958f82dc385a03b1471"
      },
      "cell_type": "code",
      "source": "x \nimport seaborn as sns\nreduced_df.dtypes",
      "execution_count": 35,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "name 'x' is not defined",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-35-263e8f129982>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mx\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mseaborn\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0msns\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mreduced_df\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdtypes\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'x' is not defined"
          ]
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "67f0dc910fa2aa6e114d119bd98c2373abfe60a2"
      },
      "cell_type": "code",
      "source": "z\ng = sns.FacetGrid(reduced_df, hue='label', size=6).map(plt.scatter, 'X', 'Y').add_legend()",
      "execution_count": 36,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "name 'z' is not defined",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-36-a2518c8f0a76>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mz\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0msns\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mFacetGrid\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mreduced_df\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mhue\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'label'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msize\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m6\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmap\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mplt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mscatter\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'X'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'Y'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0madd_legend\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'z' is not defined"
          ]
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "c1636cc5866e6a234c4cc35febc29dc20c362c3e"
      },
      "cell_type": "code",
      "source": "Converted 2 dimensions are X and Y and visualization is shown in 2-D. Some more points anout t-sne:\n\nPerplexity ensures the number of neighbours t-sne is preserving on the basis of distance metric.\nTry perplexity value in the range of 5 to 50, as suggested in research paper.\nTry multiple value of steps untill changes in cluster is saturated.\nDistance between clusters may not be useful.\nWith varying value of perplexity and steps, we may see some shapes of cluster, these are just meaningless.\nt-sne generally expands dense cluster and shrink sparse cluster. Read amazing blog on interpretation of t-sne results here",
      "execution_count": 37,
      "outputs": [
        {
          "output_type": "error",
          "ename": "SyntaxError",
          "evalue": "invalid syntax (<ipython-input-37-ee75f2f9c2b8>, line 1)",
          "traceback": [
            "\u001b[0;36m  File \u001b[0;32m\"<ipython-input-37-ee75f2f9c2b8>\"\u001b[0;36m, line \u001b[0;32m1\u001b[0m\n\u001b[0;31m    Converted 2 dimensions are X and Y and visualization is shown in 2-D. Some more points anout t-sne:\u001b[0m\n\u001b[0m              ^\u001b[0m\n\u001b[0;31mSyntaxError\u001b[0m\u001b[0;31m:\u001b[0m invalid syntax\n"
          ]
        }
      ]
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "fb044082276aa69698e3a617d70bfb79e1ad5ee5"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "eb374cdabd889742b9ebdb93dcdf0e36ede8ffd4"
      },
      "cell_type": "markdown",
      "source": "**References**\n1. https://www.kaggle.com/saurabh105/t-sne-visualization-on-mnist-dataset\n2. Wattenberg, et al., \"How to Use t-SNE Effectively\", Distill, 2016. http://doi.org/10.23915/distill.00002\n3. https://www.kaggle.com/accepteddoge/fashion-mnist-with-numpy-neural-network"
    },
    {
      "metadata": {
        "trusted": true,
        "_uuid": "578e685729d72d10ecbed6752b324c30c8e92c72"
      },
      "cell_type": "code",
      "source": "",
      "execution_count": null,
      "outputs": []
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3.6.6",
      "mimetype": "text/x-python",
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "pygments_lexer": "ipython3",
      "nbconvert_exporter": "python",
      "file_extension": ".py"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 1
}