TF 6 Learning The TensorFlow Way of k-means.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def create_samples(n_clusters, n_samples_per_cluster, n_features, embiggen_factor, seed):\n",
    "    np.random.seed(seed)\n",
    "    slices = []\n",
    "    centroids = []\n",
    "    \n",
    "    # Create samples for each cluster\n",
    "    for i in range(n_clusters):\n",
    "        samples = tf.random_normal((n_samples_per_cluster, n_features),\n",
    "                                   mean=0.0, stddev=5.0, dtype=tf.float32, seed=seed, name=\"cluster_{}\".format(i))\n",
    "        \n",
    "        current_centroid = (np.random.random((1, n_features)) * embiggen_factor) - (embiggen_factor/2)\n",
    "        \n",
    "        centroids.append(current_centroid)\n",
    "        \n",
    "        samples += current_centroid\n",
    "        \n",
    "        slices.append(samples)\n",
    "        \n",
    "    # Create a big \"samples\" dataset\n",
    "    samples = tf.concat(slices, 0, name='samples')\n",
    "    centroids = tf.concat(centroids, 0, name='centroids')\n",
    "    return centroids, samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "def plot_clusters(all_samples, centroids, n_samples_per_cluster):\n",
    "     \n",
    "    #Plot out the different clusters\n",
    "     #Choose a different colour for each cluster\n",
    "    colour = plt.cm.rainbow(np.linspace(0,1,len(centroids)))\n",
    "    for i, centroid in enumerate(centroids):\n",
    "        samples = all_samples[i*n_samples_per_cluster:(i+1)*n_samples_per_cluster]\n",
    "        plt.scatter(samples[:,0], samples[:,1], c=colour[i])\n",
    "         #Grab just the samples fpr the given cluster and plot them out with a new colour\n",
    "         #Also plot centroid\n",
    "        plt.plot(centroid[0], centroid[1], markersize=35, marker=\"x\", color='k', mew=10)\n",
    "        plt.plot(centroid[0], centroid[1], markersize=30, marker=\"x\", color='m', mew=5)\n",
    "    plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "n_features = 2\n",
    "n_clusters = 3\n",
    "n_samples_per_cluster = 500\n",
    "seed = 700\n",
    "embiggen_factor = 70\n",
    "\n",
    "centroids, samples = create_samples(n_clusters, n_samples_per_cluster, n_features, embiggen_factor, seed)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = tf.global_variables_initializer()\n",
    "with tf.Session() as session:\n",
    "    sample_values = session.run(samples)\n",
    "    centroid_values = session.run(centroids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD8CAYAAAB0IB+mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nj8fj6RC84Hs8Hk+H8P8DQsKFk80ovyYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1f0e0cafeb8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_clusters(sample_values, centroid_values, n_samples_per_cluster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def choose_random_centroids(samples, n_clusters):\n",
    "    # Step 0: Initialisation: Select `n_clusters` number of random points\n",
    "    n_samples = tf.shape(samples)[0]\n",
    "    print(n_samples)\n",
    "    random_indices = tf.random_shuffle(tf.range(0, n_samples))\n",
    "    begin = [0,]\n",
    "    size = [n_clusters,]\n",
    "    #size[0] = n_clusters\n",
    "    centroid_indices = tf.slice(random_indices, begin, size)\n",
    "    initial_centroids = tf.gather(samples, centroid_indices)\n",
    "    return initial_centroids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"strided_slice_1:0\", shape=(), dtype=int32)\n"
     ]
    }
   ],
   "source": [
    "initial_centroids = choose_random_centroids(samples, n_clusters)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "with tf.Session() as session:\n",
    "    updated_centroid_value = session.run(initial_centroids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-12.57284737,  16.90243149],\n",
       "       [ -3.82149887,  16.00059319],\n",
       "       [-36.63245392,  12.81079769]], dtype=float32)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "updated_centroid_value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-24.23473167,  -5.63619328],\n",
       "       [-21.30787468,   0.05049753],\n",
       "       [-29.07078743, -15.99440384],\n",
       "       ..., \n",
       "       [-19.84384918,   8.68898773],\n",
       "       [-12.37724876,  19.23770142],\n",
       "       [-21.83363724,  21.7114563 ]], dtype=float32)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD8CAYAAAB0IB+mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/x97ZKdm2zbilwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1f0df425fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_clusters(sample_values, updated_centroid_value, n_samples_per_cluster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "500"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_samples_per_cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"strided_slice_2:0\", shape=(), dtype=int32)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD8CAYAAAB0IB+mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ncILvcDgcbcL/D6SA6EcIH6J/AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1f0e10795c0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "\n",
    "n_features = 2\n",
    "n_clusters = 3\n",
    "n_samples_per_cluster = 500\n",
    "seed = 700\n",
    "embiggen_factor = 70\n",
    "\n",
    "centroids, samples = create_samples(n_clusters, n_samples_per_cluster, n_features, embiggen_factor, seed)\n",
    "initial_centroids = choose_random_centroids(samples, n_clusters)\n",
    "\n",
    "model = tf.global_variables_initializer()\n",
    "with tf.Session() as session:\n",
    "    sample_values = session.run(samples)\n",
    "    updated_centroid_value = session.run(initial_centroids)\n",
    "\n",
    "plot_clusters(sample_values, updated_centroid_value, n_samples_per_cluster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def assign_to_nearest(samples, centroids):\n",
    "\n",
    "    expanded_vectors = tf.expand_dims(samples, 0)\n",
    "    expanded_centroids = tf.expand_dims(centroids, 1)\n",
    "    distances = tf.reduce_sum( tf.square(\n",
    "               tf.subtract(expanded_vectors, expanded_centroids)), 2)\n",
    "    mins = tf.argmin(distances, 0)\n",
    "    nearest_indices = mins\n",
    "    return nearest_indices\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_centroids(samples, nearest_indices, n_clusters):\n",
    "    \n",
    "    # Updates the centroid to be the mean of all samples associated with it.\n",
    "    nearest_indices = tf.to_int32(nearest_indices)\n",
    "    \n",
    "    partitions = tf.dynamic_partition(samples, nearest_indices, n_clusters)\n",
    "    \n",
    "    new_centroids = tf.concat([tf.expand_dims(tf.reduce_mean(partition, 0), 0) for partition in partitions], 0)\n",
    "    return new_centroids"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1500, 2)"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sample_values.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3, 2)"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "centroid_values.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(3, 1, 2)"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "1,1500,2\n",
    "3,1,2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "a32 = np.array([ [1,2],\n",
    "           [3,4],\n",
    "           [5,6]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "b22 = np.array([ [1,2],\n",
    "           [5,6]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "a32n = a32.reshape((3,1,2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "b22n = b22.reshape((2,1,2)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "d = np.subtract(a32n,b22n)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2, 3, 2)"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0,  4,  8],\n",
       "       [-8, -4,  0]])"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d.sum(axis=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[ 0,  0],\n",
       "        [ 2,  2],\n",
       "        [ 4,  4]],\n",
       "\n",
       "       [[-4, -4],\n",
       "        [-2, -2],\n",
       "        [ 0,  0]]])"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Help on function dynamic_partition in module tensorflow.python.ops.gen_data_flow_ops:\n",
      "\n",
      "dynamic_partition(data, partitions, num_partitions, name=None)\n",
      "    Partitions `data` into `num_partitions` tensors using indices from `partitions`.\n",
      "    \n",
      "    For each index tuple `js` of size `partitions.ndim`, the slice `data[js, ...]`\n",
      "    becomes part of `outputs[partitions[js]]`.  The slices with `partitions[js] = i`\n",
      "    are placed in `outputs[i]` in lexicographic order of `js`, and the first\n",
      "    dimension of `outputs[i]` is the number of entries in `partitions` equal to `i`.\n",
      "    In detail,\n",
      "    \n",
      "    ```python\n",
      "        outputs[i].shape = [sum(partitions == i)] + data.shape[partitions.ndim:]\n",
      "    \n",
      "        outputs[i] = pack([data[js, ...] for js if partitions[js] == i])\n",
      "    ```\n",
      "    \n",
      "    `data.shape` must start with `partitions.shape`.\n",
      "    \n",
      "    For example:\n",
      "    \n",
      "    ```python\n",
      "        # Scalar partitions.\n",
      "        partitions = 1\n",
      "        num_partitions = 2\n",
      "        data = [10, 20]\n",
      "        outputs[0] = []  # Empty with shape [0, 2]\n",
      "        outputs[1] = [[10, 20]]\n",
      "    \n",
      "        # Vector partitions.\n",
      "        partitions = [0, 0, 1, 1, 0]\n",
      "        num_partitions = 2\n",
      "        data = [10, 20, 30, 40, 50]\n",
      "        outputs[0] = [10, 20, 50]\n",
      "        outputs[1] = [30, 40]\n",
      "    ```\n",
      "    \n",
      "    See `dynamic_stitch` for an example on how to merge partitions back.\n",
      "    \n",
      "    <div style=\"width:70%; margin:auto; margin-bottom:10px; margin-top:20px;\">\n",
      "    <img style=\"width:100%\" src=\"https://www.tensorflow.org/images/DynamicPartition.png\" alt>\n",
      "    </div>\n",
      "    \n",
      "    Args:\n",
      "      data: A `Tensor`.\n",
      "      partitions: A `Tensor` of type `int32`.\n",
      "        Any shape.  Indices in the range `[0, num_partitions)`.\n",
      "      num_partitions: An `int` that is `>= 1`.\n",
      "        The number of partitions to output.\n",
      "      name: A name for the operation (optional).\n",
      "    \n",
      "    Returns:\n",
      "      A list of `num_partitions` `Tensor` objects with the same type as `data`.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "help(tf.dynamic_partition)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Tensor(\"strided_slice_6:0\", shape=(), dtype=int32)\n",
      "[[-23.41298866  -5.0403614 ]\n",
      " [-12.4037981   16.87123489]\n",
      " [-29.43318748  24.97486687]]\n",
      "[[-31.39747238  25.7439785 ]\n",
      " [-14.41479874  17.914711  ]\n",
      " [-24.26021194  -5.22723341]]\n",
      "[[-20.26961517   0.14267336]\n",
      " [-28.45182228 -10.16779709]\n",
      " [-23.2160778   23.42747879]]\n",
      "[[-25.3943634   -5.88413334]\n",
      " [-21.75144577  21.31660652]\n",
      " [-16.07864761  -4.77323008]]\n",
      "[[-10.22910404  20.33134079]\n",
      " [-23.83283615  -6.07269669]\n",
      " [-25.01178169  21.28966713]]\n",
      "[[-20.73738861  23.02637863]\n",
      " [-21.29577827  -2.41832829]\n",
      " [-34.27545929  21.10885239]]\n",
      "[[-18.89899635   6.71370983]\n",
      " [-31.39985275  21.47494316]\n",
      " [-32.41139984  30.53956032]]\n",
      "[[-35.29745865  13.78770924]\n",
      " [-23.49595642  -5.36094475]\n",
      " [-21.8049984   21.80998611]]\n",
      "[[-17.27680588  19.11837387]\n",
      " [-33.45402145  25.73194695]\n",
      " [-23.82203865  -6.17243052]]\n",
      "[[-12.81673813  14.31089401]\n",
      " [-28.69512939  24.02179337]\n",
      " [-24.67459869  -6.89458084]]\n",
      "[[-19.09632492  24.87194633]\n",
      " [-24.33684349  19.08260918]\n",
      " [-23.65022659  -4.95297861]]\n",
      "[[-19.5936451    2.60443211]\n",
      " [-31.23334122  25.75405312]\n",
      " [-16.77768326  23.31212997]]\n",
      "[[-30.40720749  25.18817139]\n",
      " [-15.24376583  13.79871941]\n",
      " [-24.46020126  -7.89788771]]\n",
      "[[-13.2857008   17.08226776]\n",
      " [-30.13661003  24.75370026]\n",
      " [-23.68984985  -6.04603815]]\n",
      "[[-26.95560265  24.35166168]\n",
      " [-24.19052315  -5.84034586]\n",
      " [-11.74774837  14.99920273]]\n",
      "[[-28.99707222   4.30789852]\n",
      " [-29.04013443 -11.6475563 ]\n",
      " [-19.8191452   17.53910065]]\n",
      "[[-29.98380852  25.28877068]\n",
      " [-22.81941986  -2.70999789]\n",
      " [-11.84523392  18.06405449]]\n",
      "[[-28.99352264  -6.42686558]\n",
      " [-21.26214027  -6.82503796]\n",
      " [-22.1149559   20.45858192]]\n",
      "[[-17.00811768   7.32246685]\n",
      " [-18.20638466  30.04694748]\n",
      " [-30.64373207  17.93546295]]\n",
      "[[-15.14836502  16.51157188]\n",
      " [-27.11360168   5.06521702]\n",
      " [-22.37810326  27.4862709 ]]\n",
      "[[-27.71869087  22.91552925]\n",
      " [-19.14962387   6.04467201]\n",
      " [-32.86002731  27.65197563]]\n",
      "[[-36.7016449   25.31736946]\n",
      " [-21.233078    22.40108681]\n",
      " [-22.05782127  -2.76248741]]\n",
      "[[-21.26861     -1.30297685]\n",
      " [-36.41263199  24.23199844]\n",
      " [-22.00313377  22.79488754]]\n",
      "[[-24.11866188  -6.09012365]\n",
      " [-29.20363998  25.27437401]\n",
      " [-13.61390209  15.71132088]]\n",
      "[[-14.73783112  10.85649014]\n",
      " [ -3.29060507  14.24812412]\n",
      " [-28.41527557  13.54333115]]\n",
      "[[-30.96854973  26.01718712]\n",
      " [-14.5755167   17.63080597]\n",
      " [-23.62232971  -5.87738657]]\n",
      "[[-14.76795292  14.07003975]\n",
      " [-24.76126289  -5.07469177]\n",
      " [-26.1525135   25.3509655 ]]\n",
      "[[-26.30688286 -11.09183025]\n",
      " [-21.86023903  21.21325111]\n",
      " [-22.67162323  -3.3063097 ]]\n",
      "[[-21.97516632  21.35658073]\n",
      " [-24.52194977 -13.34427929]\n",
      " [-23.63417816  -3.87993002]]\n",
      "[[-26.79693031 -12.75020123]\n",
      " [-23.39624214  -4.38699341]\n",
      " [-22.09840012  21.02902985]]\n",
      "[[-23.30913925  -4.95942259]\n",
      " [-27.33899498  23.50531197]\n",
      " [-10.44200134  17.30614281]]\n",
      "[[-21.82613945  -6.05653477]\n",
      " [-31.92991829  16.06367683]\n",
      " [-17.38480377  20.8972435 ]]\n",
      "[[-27.2111702   27.85228348]\n",
      " [-32.97491837  22.781353  ]\n",
      " [-18.79349709   5.04346371]]\n",
      "[[-28.43749046  -1.4650315 ]\n",
      " [-21.86543846  20.7305851 ]\n",
      " [-22.63897133  -8.03948879]]\n",
      "[[-32.82534027  27.97127533]\n",
      " [-18.42123985   5.18195915]\n",
      " [-27.4802742   18.87075806]]\n",
      "[[-22.45156288  -6.67368174]\n",
      " [-22.39061737  20.89982986]\n",
      " [-30.94618797  -3.52983761]]\n",
      "[[-24.05086136  -5.83337927]\n",
      " [-13.66450691  16.92757607]\n",
      " [-30.4299202   25.50379372]]\n",
      "[[-28.51125145  14.60270786]\n",
      " [-18.96606255  21.75871468]\n",
      " [-23.80223465  -7.62687206]]\n",
      "[[-13.64116955  17.56087685]\n",
      " [-30.54552078  25.75187874]\n",
      " [-24.07309151  -5.04134417]]\n",
      "[[-25.19698906 -14.07178211]\n",
      " [-21.99043083  21.47822189]\n",
      " [-25.47599792  -2.8230257 ]]\n",
      "[[-31.48352623  27.18567467]\n",
      " [-18.40654564   6.36221981]\n",
      " [-30.25892067   9.03251457]]\n",
      "[[-21.35606575  -7.55996513]\n",
      " [-22.53646088  17.9753685 ]\n",
      " [-27.5132103  -12.45115185]]\n",
      "[[-14.31371784  12.3809042 ]\n",
      " [-26.0713501   24.43587303]\n",
      " [-24.27007675  -6.68309975]]\n",
      "[[-32.66195297  26.829813  ]\n",
      " [-21.03496933  21.95072937]\n",
      " [-19.68479729   2.03490114]]\n",
      "[[-13.62142658  17.4877243 ]\n",
      " [-30.60780716  25.71541023]\n",
      " [-23.94958687  -5.29754353]]\n",
      "[[-15.69008636  14.58912945]\n",
      " [-26.49517441  -2.55867958]\n",
      " [-27.49112129  27.47933388]]\n",
      "[[-13.78162575  26.20346451]\n",
      " [-30.81178856  21.52820587]\n",
      " [-18.25449371   4.93637276]]\n",
      "[[-13.67583942  16.74686623]\n",
      " [-23.93459511  -5.64014673]\n",
      " [-30.22063255  25.60948372]]\n",
      "[[-25.01699257  -5.20159101]\n",
      " [-13.52454567  14.49045849]\n",
      " [-29.12387657  26.07470512]]\n",
      "[[-12.13601494   6.70997953]\n",
      " [-25.3228569   12.30826283]\n",
      " [ -8.17657757  21.14234161]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAD8CAYAAAB0IB+mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1f0e0e387b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "n_features = 2\n",
    "n_clusters = 3\n",
    "n_samples_per_cluster = 500\n",
    "seed = 700\n",
    "embiggen_factor = 70\n",
    "\n",
    "\n",
    "data_centroids, samples = create_samples(n_clusters, n_samples_per_cluster, n_features, embiggen_factor, seed)\n",
    "initial_centroids = choose_random_centroids(samples, n_clusters)\n",
    "nearest_indices = assign_to_nearest(samples, initial_centroids)\n",
    "updated_centroids = update_centroids(samples, nearest_indices, n_clusters)\n",
    "\n",
    "model = tf.global_variables_initializer()\n",
    "with tf.Session() as session:\n",
    "    sample_values = session.run(samples)\n",
    "    \n",
    "    for i in range(50):\n",
    "        \n",
    "        updated_centroid_value = session.run(updated_centroids)\n",
    "        print(updated_centroid_value)\n",
    "\n",
    "plot_clusters(sample_values, updated_centroid_value, n_samples_per_cluster)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}