P4DS4D; 15; Clustering.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:5f0574628bde2f17832893fb06b2295f6f5a3ce5c77c9b533fa3188492cf65d9"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "markdown",
     "metadata": {},
     "source": [
      "http://www.slideshare.net/SarahGuido/kmeans-clustering-with-scikitlearn<BR>\n",
      "http://scikit-learn.org/stable/auto_examples/cluster/plot_kmeans_digits.html#example-cluster-plot-kmeans-digits-py<BR>\n",
      "http://scikit-learn.org/stable/auto_examples/document_clustering.html#example-document-clustering-py<BR>\n",
      "https://www.mail-archive.com/scikit-learn-general@lists.sourceforge.net/msg08695.html<BR>\n",
      "http://en.wikipedia.org/wiki/Cluster_analysis<BR>\n",
      "http://docs.scipy.org/doc/scipy-0.14.0/reference/cluster.html<BR>\n",
      "http://scikit-learn.org/stable/modules/clustering.html<BR>\n",
      "http://scikit-learn.org/stable/modules/classes.html#module-sklearn.cluster<BR>\n",
      "http://scikit-learn.org/stable/modules/generated/sklearn.cluster.AgglomerativeClustering.html#sklearn.cluster.AgglomerativeClustering<BR>\n",
      "http://scikit-learn.org/stable/auto_examples/cluster/plot_agglomerative_clustering_metrics.html#example-cluster-plot-agglomerative-clustering-metrics-py<BR>\n",
      "http://scikit-learn.org/stable/auto_examples/cluster/plot_digits_linkage.html#example-cluster-plot-digits-linkage-py<BR>\n",
      "http://people.cs.uchicago.edu/~dinoj/manifold/swissroll.html<BR>"
     ]
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "K-means"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.datasets import load_digits\n",
      "digits = load_digits()\n",
      "X = digits.data\n",
      "ground_truth = digits.target\n",
      "\n",
      "#print digits.DESCR"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 1
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.decomposition import PCA\n",
      "from sklearn.preprocessing import scale\n",
      "pca = PCA(n_components=40)\n",
      "Cx = pca.fit_transform(scale(X))\n",
      "print 'Explained variance %0.3f' % sum(pca.explained_variance_ratio_)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Explained variance 0.951\n"
       ]
      }
     ],
     "prompt_number": 2
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.cluster import KMeans\n",
      "clustering = KMeans(n_clusters=10, n_init=10, random_state=1)\n",
      "clustering.fit(Cx)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 3,
       "text": [
        "KMeans(copy_x=True, init='k-means++', max_iter=300, n_clusters=10, n_init=10,\n",
        "    n_jobs=1, precompute_distances=True, random_state=1, tol=0.0001,\n",
        "    verbose=0)"
       ]
      }
     ],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import numpy as np\n",
      "import pandas as pd\n",
      "ms = np.column_stack((ground_truth,clustering.labels_))\n",
      "df = pd.DataFrame(ms, columns = ['Ground truth','Clusters'])\n",
      "pd.crosstab(df['Ground truth'], df['Clusters'], margins=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
        "<table border=\"1\" class=\"dataframe\">\n",
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th>Clusters</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>All</th>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Ground truth</th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "    </tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>   2</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 176</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  178</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
        "      <td> 107</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>  47</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 27</td>\n",
        "      <td>   1</td>\n",
        "      <td>  182</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
        "      <td>  20</td>\n",
        "      <td>   6</td>\n",
        "      <td>   0</td>\n",
        "      <td> 1</td>\n",
        "      <td>   3</td>\n",
        "      <td>  0</td>\n",
        "      <td>   3</td>\n",
        "      <td>   0</td>\n",
        "      <td> 45</td>\n",
        "      <td>  99</td>\n",
        "      <td>  177</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
        "      <td>   6</td>\n",
        "      <td> 155</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  11</td>\n",
        "      <td>   0</td>\n",
        "      <td>  1</td>\n",
        "      <td>  10</td>\n",
        "      <td>  183</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 3</td>\n",
        "      <td> 154</td>\n",
        "      <td>  9</td>\n",
        "      <td>  15</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  181</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>5</th>\n",
        "      <td>   6</td>\n",
        "      <td>  74</td>\n",
        "      <td>   2</td>\n",
        "      <td> 0</td>\n",
        "      <td>   4</td>\n",
        "      <td>  0</td>\n",
        "      <td>  48</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  48</td>\n",
        "      <td>  182</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>6</th>\n",
        "      <td>   6</td>\n",
        "      <td>   0</td>\n",
        "      <td> 174</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>  181</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>7</th>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>   3</td>\n",
        "      <td> 17</td>\n",
        "      <td> 150</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>   8</td>\n",
        "      <td>  179</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>8</th>\n",
        "      <td>  90</td>\n",
        "      <td>  51</td>\n",
        "      <td>   2</td>\n",
        "      <td> 0</td>\n",
        "      <td>   3</td>\n",
        "      <td>  0</td>\n",
        "      <td>  15</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  13</td>\n",
        "      <td>  174</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>9</th>\n",
        "      <td>   1</td>\n",
        "      <td> 146</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>  14</td>\n",
        "      <td>  4</td>\n",
        "      <td>  14</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>   1</td>\n",
        "      <td>  180</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>All</th>\n",
        "      <td> 237</td>\n",
        "      <td> 432</td>\n",
        "      <td> 178</td>\n",
        "      <td> 4</td>\n",
        "      <td> 230</td>\n",
        "      <td> 30</td>\n",
        "      <td> 256</td>\n",
        "      <td> 177</td>\n",
        "      <td> 73</td>\n",
        "      <td> 180</td>\n",
        "      <td> 1797</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "<p>11 rows \u00d7 11 columns</p>\n",
        "</div>"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 4,
       "text": [
        "Clusters        0    1    2  3    4   5    6    7   8    9   All\n",
        "Ground truth                                                    \n",
        "0               0    0    0  0    2   0    0  176   0    0   178\n",
        "1             107    0    0  0   47   0    0    0  27    1   182\n",
        "2              20    6    0  1    3   0    3    0  45   99   177\n",
        "3               6  155    0  0    0   0   11    0   1   10   183\n",
        "4               0    0    0  3  154   9   15    0   0    0   181\n",
        "5               6   74    2  0    4   0   48    0   0   48   182\n",
        "6               6    0  174  0    0   0    0    1   0    0   181\n",
        "7               1    0    0  0    3  17  150    0   0    8   179\n",
        "8              90   51    2  0    3   0   15    0   0   13   174\n",
        "9               1  146    0  0   14   4   14    0   0    1   180\n",
        "All           237  432  178  4  230  30  256  177  73  180  1797\n",
        "\n",
        "[11 rows x 11 columns]"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# How good is the results?\n",
      "from sklearn import metrics\n",
      "metrics.adjusted_rand_score(ground_truth,clustering.labels_)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 5,
       "text": [
        "0.51463943542053603"
       ]
      }
     ],
     "prompt_number": 5
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "dist = clustering.fit_transform(Cx)\n",
      "print np.min(dist,axis=0)\n",
      "print np.argmin(dist,axis=0)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "[  2.63928052   3.27313242   2.32088881  20.12704663   3.51263982\n",
        "   4.25806578   3.67317681   2.15597327   4.12493572   3.67969141]\n",
        "[1327  448  272  873   97  467  983  682  979  833]\n"
       ]
      }
     ],
     "prompt_number": 6
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import matplotlib.pyplot as plt\n",
      "for k,img in enumerate(np.argmin(dist,axis=0)):\n",
      "    cluster = clustering.labels_[img]\n",
      "    plt.subplot(2, 5, cluster)\n",
      "    plt.imshow(digits.images[img],cmap='binary',interpolation='none')\n",
      "    plt.title('cl '+str(cluster))\n",
      "plt.show()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAWYAAADeCAYAAAAOyKq6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\njx7g8QAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x17ceacf8>"
       ]
      }
     ],
     "prompt_number": 7
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "inertia = list()\n",
      "delta_inertia = list()\n",
      "for k in range(1,21):\n",
      "    clustering = KMeans(n_clusters=k, n_init=10, random_state=1)\n",
      "    clustering.fit(Cx)\n",
      "    if inertia:\n",
      "        delta_inertia.append(inertia[-1] - clustering.inertia_)\n",
      "    inertia.append(clustering.inertia_)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "2 9189.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "3 6581.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "4 4721.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "5 4010.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "6 3723.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "7 3245.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "8 2981.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "9 3127.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "10 2460.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "11 2183.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "12 2326.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "13 2027.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "14 875.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "15 1563.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "16 1596.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "17 1959.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "18 978.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "19 1151.0 "
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "20 935.0\n"
       ]
      }
     ],
     "prompt_number": 16
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "len([k for k in range(1,21)])"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 13,
       "text": [
        "19"
       ]
      }
     ],
     "prompt_number": 13
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import matplotlib.pyplot as plt\n",
      "plt.figure()\n",
      "plt.plot([k for k in range(2,21)], delta_inertia, 'ko-')\n",
      "plt.xlabel('Number of clusters')\n",
      "plt.ylabel('Rate of change of inertia')\n",
      "plt.show()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAZMAAAEPCAYAAACHuClZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x18469b70>"
       ]
      }
     ],
     "prompt_number": 24
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Mini-batch K-means"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.cluster import MiniBatchKMeans\n",
      "batch_clustering = MiniBatchKMeans(n_clusters=10, random_state=1)\n",
      "\n",
      "batch = 100\n",
      "guessed_labels = list()\n",
      "inertia = 0\n",
      "for row in range(0,len(Cx),batch):\n",
      "    if row+batch < len(Cx):\n",
      "        feed = Cx[row:row+batch,:]\n",
      "    else:\n",
      "        feed = Cx[row:,:]\n",
      "    batch_clustering.partial_fit(feed)\n",
      "    # We have to stack results in a list, because MiniBatchKMean\n",
      "    # does not take track of all the batches\n",
      "    guessed_labels.append(batch_clustering.labels_)\n",
      "    inertia += batch_clustering.inertia_\n",
      "\n",
      "# NumPy hstack turns a list of arrays into an array\n",
      "# by inspecting the variable guess_labels you can know \n",
      "# the assigned cluster\n",
      "guessed_labels = np.hstack(guessed_labels)\n",
      "\n",
      "print \"Kmeans inertia: %0.1f\\nMiniBatchKmeans inertia: %0.1f\" % (clustering.inertia_,inertia)\n"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Kmeans inertia: 48591.7\n",
        "MiniBatchKmeans inertia: 67027.5\n"
       ]
      }
     ],
     "prompt_number": 51
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# How good is the results?\n",
      "from sklearn import metrics\n",
      "metrics.adjusted_rand_score(ground_truth,guessed_labels)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 100,
       "text": [
        "0.57337583103264378"
       ]
      }
     ],
     "prompt_number": 100
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Hierarchical clustering"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.cluster import AgglomerativeClustering\n",
      "# Affinity = {\u201ceuclidean\u201d, \u201cl1\u201d, \u201cl2\u201d, \u201cmanhattan\u201d, \u201ccosine\u201d}\n",
      "# Linkage = {\u201cward\u201d, \u201ccomplete\u201d, \u201caverage\u201d}\n",
      "Hclustering = AgglomerativeClustering(n_clusters=10, affinity='euclidean', linkage='ward')\n",
      "Hclustering.fit(Cx)\n",
      "\n",
      "ms = np.column_stack((ground_truth,Hclustering.labels_))\n",
      "df = pd.DataFrame(ms, columns = ['Ground truth','Clusters'])\n",
      "pd.crosstab(df['Ground truth'], df['Clusters'], margins=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
        "<table border=\"1\" class=\"dataframe\">\n",
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th>Clusters</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>All</th>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Ground truth</th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "    </tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 178</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  178</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
        "      <td>   1</td>\n",
        "      <td>  27</td>\n",
        "      <td>  61</td>\n",
        "      <td> 0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  93</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  182</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
        "      <td>   0</td>\n",
        "      <td> 129</td>\n",
        "      <td>   0</td>\n",
        "      <td> 1</td>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  46</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  177</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
        "      <td>   0</td>\n",
        "      <td>   4</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td> 160</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  19</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  183</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td> 165</td>\n",
        "      <td> 1</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   4</td>\n",
        "      <td> 10</td>\n",
        "      <td>   0</td>\n",
        "      <td>  181</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>5</th>\n",
        "      <td> 148</td>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td> 0</td>\n",
        "      <td>  31</td>\n",
        "      <td>   0</td>\n",
        "      <td>   1</td>\n",
        "      <td>   1</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  182</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>6</th>\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> 180</td>\n",
        "      <td>   1</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  181</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>7</th>\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>   1</td>\n",
        "      <td>  8</td>\n",
        "      <td> 169</td>\n",
        "      <td>  179</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>8</th>\n",
        "      <td>   0</td>\n",
        "      <td>   3</td>\n",
        "      <td>   1</td>\n",
        "      <td> 0</td>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 169</td>\n",
        "      <td>  0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  174</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>9</th>\n",
        "      <td>   2</td>\n",
        "      <td>   0</td>\n",
        "      <td>  15</td>\n",
        "      <td> 0</td>\n",
        "      <td> 133</td>\n",
        "      <td>   0</td>\n",
        "      <td>   1</td>\n",
        "      <td>  26</td>\n",
        "      <td>  3</td>\n",
        "      <td>   0</td>\n",
        "      <td>  180</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>All</th>\n",
        "      <td> 153</td>\n",
        "      <td> 164</td>\n",
        "      <td> 242</td>\n",
        "      <td> 2</td>\n",
        "      <td> 326</td>\n",
        "      <td> 178</td>\n",
        "      <td> 182</td>\n",
        "      <td> 360</td>\n",
        "      <td> 21</td>\n",
        "      <td> 169</td>\n",
        "      <td> 1797</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "<p>11 rows \u00d7 11 columns</p>\n",
        "</div>"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 97,
       "text": [
        "Clusters        0    1    2  3    4    5    6    7   8    9   All\n",
        "Ground truth                                                     \n",
        "0               0    0    0  0    0  178    0    0   0    0   178\n",
        "1               1   27   61  0    0    0    0   93   0    0   182\n",
        "2               0  129    0  1    1    0    0   46   0    0   177\n",
        "3               0    4    0  0  160    0    0   19   0    0   183\n",
        "4               1    0  165  1    0    0    0    4  10    0   181\n",
        "5             148    1    0  0   31    0    1    1   0    0   182\n",
        "6               0    0    0  0    0    0  180    1   0    0   181\n",
        "7               1    0    0  0    0    0    0    1   8  169   179\n",
        "8               0    3    1  0    1    0    0  169   0    0   174\n",
        "9               2    0   15  0  133    0    1   26   3    0   180\n",
        "All           153  164  242  2  326  178  182  360  21  169  1797\n",
        "\n",
        "[11 rows x 11 columns]"
       ]
      }
     ],
     "prompt_number": 97
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "# How good is the results?\n",
      "from sklearn import metrics\n",
      "metrics.adjusted_rand_score(ground_truth,Hclustering.labels_)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 101,
       "text": [
        "0.64866268144435724"
       ]
      }
     ],
     "prompt_number": 101
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "DBScan"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.cluster import DBSCAN\n",
      "from collections import Counter\n",
      "for e in range(10,50, 1):\n",
      "    l = float(e)/10.\n",
      "    DB = DBSCAN(eps=l, min_samples=25, random_state=1)\n",
      "    DB.fit(Cx)\n",
      "    print l, unique(DB.labels_)\n",
      "    print Counter(DB.labels_)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "1.0 [-1]\n",
        "Counter({-1: 1797})\n",
        "1.1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1]\n",
        "Counter({-1: 1797})\n",
        "1.2"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1]\n",
        "Counter({-1: 1797})\n",
        "1.3"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1]\n",
        "Counter({-1: 1797})\n",
        "1.4"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1]\n",
        "Counter({-1: 1797})\n",
        "1.5"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1]\n",
        "Counter({-1: 1797})\n",
        "1.6"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0]\n",
        "Counter({-1: 1790, 0: 7})\n",
        "1.7"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1]\n",
        "Counter({-1: 1782, 1: 8, 0: 7})\n",
        "1.8"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1]\n",
        "Counter({-1: 1780, 1: 9, 0: 8})\n",
        "1.9"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0]\n",
        "Counter({-1: 1766, 0: 31})\n",
        "2.0"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0]\n",
        "Counter({-1: 1765, 0: 32})\n",
        "2.1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1]\n",
        "Counter({-1: 1756, 0: 35, 1: 6})\n",
        "2.2"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2]\n",
        "Counter({-1: 1749, 0: 37, 2: 6, 1: 5})\n",
        "2.3"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2]\n",
        "Counter({-1: 1744, 0: 40, 2: 8, 1: 5})\n",
        "2.4"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3]\n",
        "Counter({-1: 1729, 0: 42, 1: 16, 2: 5, 3: 5})\n",
        "2.5"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6]\n",
        "Counter({-1: 1691, 0: 44, 2: 20, 1: 15, 3: 12, 4: 6, 5: 5, 6: 4})\n",
        "2.6"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13]\n",
        "Counter({-1: 1617, 0: 45, 4: 32, 1: 27, 5: 20, 3: 10, 7: 7, 8: 6, 2: 5, 6: 5, 9: 5, 10: 5, 11: 5, 13: 5, 12: 3})\n",
        "2.7"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14]\n",
        "Counter({-1: 1546, 1: 75, 0: 50, 6: 42, 3: 15, 4: 11, 7: 10, 2: 6, 8: 6, 12: 6, 5: 5, 9: 5, 10: 5, 11: 5, 13: 5, 14: 5})\n",
        "2.8"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13]\n",
        "Counter({-1: 1491, 1: 85, 6: 59, 0: 52, 7: 30, 3: 22, 4: 12, 2: 7, 11: 7, 12: 7, 5: 6, 8: 5, 9: 5, 13: 5, 10: 4})\n",
        "2.9"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16]\n",
        "Counter({-1: 1414, 2: 110, 1: 97, 0: 60, 5: 29, 6: 15, 4: 9, 14: 8, 12: 7, 15: 7, 3: 6, 7: 6, 8: 5, 10: 5, 11: 5, 13: 5, 16: 5, 9: 4})\n",
        "3.0"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21]\n",
        "Counter({-1: 1344, 2: 116, 1: 108, 0: 63, 5: 32, 6: 16, 3: 15, 13: 10, 15: 10, 4: 9, 18: 8, 12: 7, 14: 7, 20: 7, 7: 6, 8: 6, 16: 6, 10: 5, 11: 5, 17: 5, 19: 5, 9: 4, 21: 3})\n",
        "3.1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n",
        " 24 25 26 27]\n",
        "Counter({-1: 1240, 4: 122, 1: 118, 0: 74, 8: 40, 2: 29, 5: 15, 13: 15, 10: 12, 17: 12, 14: 11, 7: 10, 15: 10, 9: 9, 3: 8, 11: 6, 12: 6, 16: 6, 18: 6, 24: 6, 6: 5, 19: 5, 20: 5, 21: 5, 22: 5, 23: 5, 25: 4, 26: 4, 27: 4})\n",
        "3.2"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21]\n",
        "Counter({-1: 1137, 1: 138, 4: 136, 0: 81, 3: 69, 2: 66, 9: 42, 5: 27, 7: 13, 10: 11, 8: 10, 6: 8, 13: 7, 11: 6, 12: 6, 18: 6, 14: 5, 15: 5, 16: 5, 17: 5, 19: 5, 21: 5, 20: 4})\n",
        "3.3"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n",
        " 24 25 26 27 28 29 30]\n",
        "Counter({-1: 1027, 1: 143, 5: 142, 0: 85, 4: 83, 2: 69, 3: 56, 6: 30, 12: 15, 9: 14, 10: 10, 7: 8, 18: 8, 19: 8, 11: 7, 17: 7, 23: 7, 13: 6, 15: 6, 22: 6, 25: 6, 27: 6, 28: 6, 8: 5, 14: 5, 16: 5, 21: 5, 26: 5, 29: 5, 20: 4, 24: 4, 30: 4})\n",
        "3.4"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n",
        " 24 25 26 27 28 29 30 31 32]\n",
        "Counter({-1: 904, 3: 152, 1: 148, 0: 94, 5: 93, 2: 80, 4: 77, 6: 35, 10: 16, 8: 15, 11: 15, 15: 15, 12: 13, 20: 12, 19: 11, 7: 10, 14: 9, 18: 8, 31: 8, 9: 7, 25: 7, 28: 7, 32: 7, 16: 6, 27: 6, 13: 5, 17: 5, 22: 5, 24: 5, 29: 5, 30: 5, 21: 4, 23: 4, 26: 4})\n",
        "3.5"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n",
        " 24 25 26 27 28 29 30 31]\n",
        "Counter({-1: 808, 1: 158, 4: 154, 0: 105, 6: 98, 5: 94, 3: 89, 15: 61, 2: 37, 7: 29, 10: 24, 8: 17, 11: 15, 14: 9, 19: 9, 9: 7, 12: 7, 17: 7, 25: 7, 16: 6, 13: 5, 18: 5, 20: 5, 23: 5, 24: 5, 26: 5, 29: 5, 21: 4, 22: 4, 27: 4, 31: 4, 28: 3, 30: 2})\n",
        "3.6"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23]\n",
        "Counter({-1: 724, 4: 179, 1: 166, 2: 154, 0: 147, 5: 113, 3: 109, 6: 61, 9: 27, 7: 22, 10: 15, 11: 8, 17: 8, 12: 7, 16: 7, 8: 6, 13: 6, 15: 6, 14: 5, 18: 5, 19: 5, 21: 5, 22: 5, 23: 4, 20: 3})\n",
        "3.7"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18 19 20 21 22 23\n",
        " 24 25]\n",
        "Counter({-1: 624, 0: 387, 1: 171, 2: 157, 3: 118, 5: 117, 4: 62, 9: 31, 10: 15, 15: 11, 8: 9, 12: 9, 16: 8, 18: 8, 6: 7, 11: 7, 13: 7, 7: 6, 14: 6, 17: 6, 25: 6, 19: 5, 20: 5, 21: 5, 23: 5, 22: 3, 24: 2})\n",
        "3.8"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13 14 15 16 17 18]\n",
        "Counter({0: 614, -1: 548, 1: 172, 4: 125, 2: 123, 3: 78, 6: 34, 8: 16, 12: 13, 11: 12, 10: 9, 5: 8, 16: 7, 7: 6, 9: 6, 18: 6, 13: 5, 14: 5, 15: 5, 17: 5})\n",
        "3.9"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11 12 13]\n",
        "Counter({0: 794, -1: 487, 1: 172, 2: 128, 3: 125, 6: 17, 7: 16, 9: 12, 4: 9, 8: 9, 10: 7, 5: 6, 11: 5, 12: 5, 13: 5})\n",
        "4.0"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6  7  8  9 10 11]\n",
        "Counter({0: 985, -1: 414, 1: 174, 2: 133, 4: 23, 5: 17, 7: 12, 6: 10, 3: 9, 8: 7, 11: 7, 9: 4, 10: 2})\n",
        "4.1"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6]\n",
        "Counter({0: 1088, -1: 351, 1: 175, 2: 136, 4: 18, 5: 17, 3: 9, 6: 3})\n",
        "4.2"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5]\n",
        "Counter({0: 1151, -1: 304, 1: 176, 2: 136, 4: 19, 3: 9, 5: 2})\n",
        "4.3"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5  6]\n",
        "Counter({0: 1362, -1: 254, 1: 139, 3: 21, 2: 9, 4: 5, 5: 4, 6: 3})\n",
        "4.4"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4]\n",
        "Counter({0: 1387, -1: 232, 1: 141, 3: 23, 2: 9, 4: 5})\n",
        "4.5"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4  5]\n",
        "Counter({0: 1402, -1: 210, 1: 141, 3: 23, 2: 11, 4: 5, 5: 5})\n",
        "4.6"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4]\n",
        "Counter({0: 1575, -1: 184, 1: 23, 2: 5, 3: 5, 4: 5})\n",
        "4.7"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4]\n",
        "Counter({0: 1594, -1: 163, 1: 25, 2: 6, 3: 5, 4: 4})\n",
        "4.8"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3  4]\n",
        "Counter({0: 1614, -1: 143, 1: 25, 2: 7, 3: 5, 4: 3})\n",
        "4.9"
       ]
      },
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        " [-1  0  1  2  3]\n",
        "Counter({0: 1630, -1: 130, 1: 25, 2: 7, 3: 5})\n"
       ]
      }
     ],
     "prompt_number": 55
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "from sklearn.cluster import DBSCAN\n",
      "DB = DBSCAN(eps=4.35, min_samples=25, random_state=1)\n",
      "DB.fit(Cx)\n",
      "\n",
      "from collections import Counter\n",
      "print Counter(DB.labels_)\n",
      "\n",
      "import matplotlib.pyplot as plt\n",
      "for k,cl in enumerate(np.unique(DB.labels_)):\n",
      "    if cl >= 0:\n",
      "        example = np.min(np.where(DB.labels_==cl))\n",
      "        plt.subplot(2, 3, k)\n",
      "        plt.imshow(digits.images[example],cmap='binary',interpolation='none')\n",
      "        plt.title('cl '+str(cl))\n",
      "plt.show()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "Counter({-1: 913, 4: 222, 1: 176, 3: 162, 0: 134, 2: 104, 5: 86})\n"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAWYAAAEHCAYAAACdjuzpAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n83/R6WTO0oG6OgAAAABJRU5ErkJggg==\n",
       "text": [
        "<matplotlib.figure.Figure at 0x18497b70>"
       ]
      }
     ],
     "prompt_number": 57
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "ms = np.column_stack((ground_truth,DB.labels_))\n",
      "df = pd.DataFrame(ms, columns = ['Ground truth','Clusters'])\n",
      "pd.crosstab(df['Ground truth'], df['Clusters'], margins=True)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "html": [
        "<div style=\"max-height:1000px;max-width:1500px;overflow:auto;\">\n",
        "<table border=\"1\" class=\"dataframe\">\n",
        "  <thead>\n",
        "    <tr style=\"text-align: right;\">\n",
        "      <th>Clusters</th>\n",
        "      <th>-1</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>All</th>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>Ground truth</th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "      <th></th>\n",
        "    </tr>\n",
        "  </thead>\n",
        "  <tbody>\n",
        "    <tr>\n",
        "      <th>0</th>\n",
        "      <td>   4</td>\n",
        "      <td>   0</td>\n",
        "      <td> 174</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  178</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>1</th>\n",
        "      <td>  89</td>\n",
        "      <td>  93</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  182</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>2</th>\n",
        "      <td> 176</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>  177</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>3</th>\n",
        "      <td>  67</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td> 115</td>\n",
        "      <td>  0</td>\n",
        "      <td>  183</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>4</th>\n",
        "      <td>  95</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 86</td>\n",
        "      <td>  181</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>5</th>\n",
        "      <td> 178</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   1</td>\n",
        "      <td>   3</td>\n",
        "      <td>  0</td>\n",
        "      <td>  182</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>6</th>\n",
        "      <td>  17</td>\n",
        "      <td>   2</td>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td> 161</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  181</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>7</th>\n",
        "      <td>  76</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td> 103</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  0</td>\n",
        "      <td>  179</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>8</th>\n",
        "      <td> 115</td>\n",
        "      <td>  38</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  21</td>\n",
        "      <td>  0</td>\n",
        "      <td>  174</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>9</th>\n",
        "      <td>  96</td>\n",
        "      <td>   0</td>\n",
        "      <td>   1</td>\n",
        "      <td>   0</td>\n",
        "      <td>   0</td>\n",
        "      <td>  83</td>\n",
        "      <td>  0</td>\n",
        "      <td>  180</td>\n",
        "    </tr>\n",
        "    <tr>\n",
        "      <th>All</th>\n",
        "      <td> 913</td>\n",
        "      <td> 134</td>\n",
        "      <td> 176</td>\n",
        "      <td> 104</td>\n",
        "      <td> 162</td>\n",
        "      <td> 222</td>\n",
        "      <td> 86</td>\n",
        "      <td> 1797</td>\n",
        "    </tr>\n",
        "  </tbody>\n",
        "</table>\n",
        "<p>11 rows \u00d7 8 columns</p>\n",
        "</div>"
       ],
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 58,
       "text": [
        "Clusters       -1    0    1    2    3    4   5   All\n",
        "Ground truth                                        \n",
        "0               4    0  174    0    0    0   0   178\n",
        "1              89   93    0    0    0    0   0   182\n",
        "2             176    1    0    0    0    0   0   177\n",
        "3              67    0    0    1    0  115   0   183\n",
        "4              95    0    0    0    0    0  86   181\n",
        "5             178    0    0    0    1    3   0   182\n",
        "6              17    2    1    0  161    0   0   181\n",
        "7              76    0    0  103    0    0   0   179\n",
        "8             115   38    0    0    0   21   0   174\n",
        "9              96    0    1    0    0   83   0   180\n",
        "All           913  134  176  104  162  222  86  1797\n",
        "\n",
        "[11 rows x 8 columns]"
       ]
      }
     ],
     "prompt_number": 58
    },
    {
     "cell_type": "heading",
     "level": 1,
     "metadata": {},
     "source": [
      "Euclidean distance easy tutorial"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import numpy as np\n",
      "A = np.array([165, 55, 70])\n",
      "B = np.array([185, 60, 30])\n",
      "\n",
      "D = (A-B)\n",
      "D = D**2\n",
      "D = np.sqrt(np.sum(D))\n",
      "\n",
      "print D"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "45.0\n"
       ]
      }
     ],
     "prompt_number": 41
    }
   ],
   "metadata": {}
  }
 ]
}