Embedding_Clustering_CRNN.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "from os.path import isfile\n",
    "import keras\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Input, Dense, Bidirectional, LSTM, Dropout, Activation, GRU\n",
    "from keras.layers import Conv2D, concatenate, MaxPooling2D, Flatten, Embedding, Lambda\n",
    "\n",
    "\n",
    "from keras.callbacks import ModelCheckpoint, TensorBoard, ReduceLROnPlateau\n",
    "from keras import backend as K\n",
    "from keras.utils import np_utils\n",
    "from keras.optimizers import Adam, RMSprop\n",
    "\n",
    "from keras import regularizers\n",
    "\n",
    "\n",
    "import librosa\n",
    "import librosa.display\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load pretrained model and test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_genres = {'Electronic':0, 'Experimental':1, 'Folk':2, 'Hip-Hop':3, \n",
    "               'Instrumental':4,'International':5, 'Pop' :6, 'Rock': 7  }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/tensorflow/python/util/deprecation.py:497: calling conv1d (from tensorflow.python.ops.nn_ops) with data_format=NHWC is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "`NHWC` for data_format is deprecated, use `NWC` instead\n"
     ]
    }
   ],
   "source": [
    "from keras.models import load_model\n",
    "\n",
    "weights_path = 'models/crnn/weights.best.h5'\n",
    "model = load_model(weights_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['arr_0', 'arr_1']\n",
      "(800, 640, 128) (800,)\n"
     ]
    }
   ],
   "source": [
    "npzfile = np.load('test_arr.npz')\n",
    "print(npzfile.files)\n",
    "X_test = npzfile['arr_0']\n",
    "y_test = npzfile['arr_1']\n",
    "print(X_test.shape, y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0 7 3.5\n"
     ]
    }
   ],
   "source": [
    "y_test -= 1\n",
    "print(np.amin(y_test), np.amax(y_test), np.mean(y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(800, 640, 128)\n"
     ]
    }
   ],
   "source": [
    "X_test_raw = librosa.core.db_to_power(X_test, ref=1.0)\n",
    "X_test = np.log(X_test_raw)\n",
    "# X_test = np.expand_dims(X_test, axis = -1)\n",
    "print(X_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Extract embeddings from the concat layer of the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "layer_name = 'dense1'\n",
    "intermediate_layer_model = Model(inputs=model.input,\n",
    "                                 outputs=model.get_layer(layer_name).output)\n",
    "intermediate_output = intermediate_layer_model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(800, 64)\n"
     ]
    }
   ],
   "source": [
    "print(intermediate_output.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# print(intermediate_output[:2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Cluster using K means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.cluster import KMeans, SpectralClustering"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3433.5354\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 2805.9912\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2601.613\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2521.6172\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2470.8357\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2456.0308\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2447.3564\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2444.024\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2441.478\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2436.269\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2433.0334\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2431.047\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2430.47\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2430.47\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3480.4316\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 3040.6243\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2871.0044\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2777.165\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2739.3896\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2708.765\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2670.1128\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2586.0664\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2501.6555\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2457.3865\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2438.1526\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2433.6816\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2430.394\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2430.0093\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 14, inertia 2429.9053\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 15, inertia 2429.9053\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3279.4639\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 2953.4065\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2784.7893\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2542.129\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2466.6584\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2454.5298\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2448.408\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2443.5298\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2437.515\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2434.2424\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2431.4932\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2430.5532\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2430.47\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2430.47\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3583.9932\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 2955.2556\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2623.9495\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2514.5466\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2484.5032\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2471.3145\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2466.1538\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2455.2578\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2444.857\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2438.298\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2435.0112\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2433.3164\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2433.3164\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3667.2295\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 3117.5955\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2915.584\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2702.885\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2574.9717\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2533.7522\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2506.4282\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2487.0466\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2475.682\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2468.0557\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2460.0862\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2452.473\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2446.5078\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2444.4302\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 14, inertia 2442.7673\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 15, inertia 2439.2458\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 16, inertia 2435.45\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 17, inertia 2432.1597\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 18, inertia 2431.6758\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 19, inertia 2431.6758\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 4230.543\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 3442.045\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 3230.7837\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 3110.5737\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 3044.385\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2999.3032\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2976.0166\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2962.4417\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2950.902\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2938.3147\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2915.969\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2827.1633\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2706.6458\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2641.3035\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 14, inertia 2609.3772\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 15, inertia 2594.8105\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 16, inertia 2575.1064\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 17, inertia 2545.0085\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 18, inertia 2511.0293\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 19, inertia 2487.1526\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 20, inertia 2468.6733\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 21, inertia 2456.9646\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 22, inertia 2444.9663\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 23, inertia 2437.8357\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 24, inertia 2435.0266\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 25, inertia 2433.4807\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 26, inertia 2433.3884\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 27, inertia 2433.3884\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3514.7786\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 2893.4492\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2639.2485\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2514.9958\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2480.4658\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2460.3823\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2449.3481\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2438.742\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2435.322\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2433.807\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2432.811\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2432.0967\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2431.9531\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2431.6462\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 14, inertia 2431.6462\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3089.1226\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 2751.2195\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2659.6963\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2620.3293\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2605.2905\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2597.7861\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2591.302\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2588.1006\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2587.291\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2586.8196\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2586.644\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2586.644\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 3655.143\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 2822.8774\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2708.562\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2656.8708\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2639.9688\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2626.3303\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2608.8875\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2595.9805\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2585.184\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2580.2227\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2576.2654\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2572.4822\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2568.9153\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2563.0251\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 14, inertia 2559.6055\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 15, inertia 2557.712\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 16, inertia 2556.2615\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 17, inertia 2555.9924\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 18, inertia 2555.9924\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n",
      "Initialization complete\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 0, inertia 4208.7954\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 1, inertia 3463.4983\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 2, inertia 2926.072\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 3, inertia 2790.6501\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 4, inertia 2715.6196\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 5, inertia 2636.495\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 6, inertia 2547.8757\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 7, inertia 2495.8013\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 8, inertia 2466.9902\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 9, inertia 2454.263\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 10, inertia 2444.0715\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 11, inertia 2435.4438\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 12, inertia 2432.87\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 13, inertia 2431.9985\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 14, inertia 2431.6462\n",
      "start iteration\n",
      "done sorting\n",
      "end inner loop\n",
      "Iteration 15, inertia 2431.6462\n",
      "center shift 0.000000e+00 within tolerance 1.690895e-05\n"
     ]
    }
   ],
   "source": [
    "kmeans = KMeans(n_clusters=8, init='random', verbose =1).fit(intermediate_output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(800,)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labels = kmeans.labels_\n",
    "labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0 1 7 7 1 1 7 3 6 0 0 0 3 1 0 7 0 3 1 1 3 1 1 5 6 1 7 1 6 1 3 6 6 6 7 3 3\n",
      " 6 0 7 3 7 3 6 6 2 6 7 3 0 3 0 3 3 3 4 3 2 7 7 7 0 5 5 4 0 0 3 0 6 7 3 3 0\n",
      " 3 0 4 0 4 6 3 0 3 0 7 3 0 0 0 3 3 4 3 3 0 3 2 3 6 3 3 5 0 5 5 0 2 5 2 1 1\n",
      " 1 1 1 1 7 6 7 6 0 1 7 3 3 0 0 3 0 6 3 0 0 3 0 0 5 0 0 7 7 6 0 6 0 4 5 2 7\n",
      " 1 1 5 0 5 0 1 3 0 2 0 7 3 1 1 1 1 7 1 2 0 5 2 3 2 3 5 1 2 1 0 3 0 3 0 0 7\n",
      " 1 0 7 1 7 1 5 7 0 1 1 5 1 4 7 4 4 4 7 7 4 7 7 7 7 7 4 7 1 7 7 4 7 7 7 7 6\n",
      " 7 7 7 3 3 2 1 7 3 3 3 3 3 7 3 3 7 3 7 0 0 3 3 7 0 1 5 5 3 1 5 7 7 7 3 7 7\n",
      " 1 1 1 7 0 7 4 7 7 7 7 7 7 7 7 7 7 0 0 0 5 1 6 3 6 3 2 0 4 1 4 2 5 2 2 3 6\n",
      " 0 3 3 3 0 0 3 3 6 4 0 3 2 6 6 6 6 1 2 4 7 5 7 7 7 4 7 7 7 7 7 7 3 7 1 1 6\n",
      " 1 4 3 3 3 0 3 3 3 3 0 1 3 1 5 1 2 2 2 2 0 2 0 2 1 3 2 4 7 3 1 2 3 3 3 1 0\n",
      " 5 1 1 2 1 1 1 5 0 1 1 1 3 2 2 6 5 0 1 2 2 2 2 0 5 1 4 4 0 5 1 5 0 1 3 6 7\n",
      " 5 5 3 3 3 0 0 6 0 0 0 5 0 1 0 5 1 0 0 5 0 1 3 3 3 3 5 1 5 5 6 5 1 5 2 1 5\n",
      " 4 6 7 7 3 6 7 0 0 3 4 6 4 4 4 4 2 2 6 0 1 1 2 2 0 5 0 0 7 7 7 0 0 5 5 5 2\n",
      " 0 4 4 0 0 4 6 1 1 6 0 0 2 0 0 4 7 2 5 5 1 1 1 1 1 1 1 1 6 1 5 5 5 5 5 5 5\n",
      " 5 5 5 6 6 6 1 6 6 5 5 5 5 5 5 5 1 5 5 5 3 5 5 3 3 3 3 6 3 6 6 1 1 5 0 5 1\n",
      " 3 3 3 5 5 1 5 5 5 5 5 5 0 0 2 7 0 6 3 6 6 3 3 3 6 6 6 6 3 0 6 0 4 2 4 5 5\n",
      " 0 6 3 4 2 4 6 6 6 2 2 2 2 2 5 3 4 7 4 4 4 4 0 2 1 7 0 7 7 7 7 2 1 0 2 0 1\n",
      " 6 2 2 2 2 2 2 4 3 6 4 3 3 2 0 0 0 0 0 0 0 0 2 4 2 4 4 6 4 2 5 3 2 4 4 4 6\n",
      " 4 4 3 6 6 6 7 2 7 7 7 5 0 6 4 4 2 2 1 4 0 6 0 4 5 1 2 2 2 7 3 3 2 6 6 6 3\n",
      " 6 6 4 6 0 1 0 1 1 1 2 5 5 5 5 5 5 0 1 0 1 0 0 1 0 1 1 1 3 3 5 5 5 5 0 1 1\n",
      " 0 0 0 1 4 5 5 5 5 5 1 2 2 4 4 3 6 6 5 5 1 5 5 4 4 4 1 5 5 5 5 5 6 5 5 0 2\n",
      " 7 2 0 0 0 0 0 0 2 1 1 3 0 5 1 5 5 1 1 1 1 1 5]\n"
     ]
    }
   ],
   "source": [
    "print(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Evaluate the output from K-means "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Adjusted Rand Index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import adjusted_rand_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.21950611088064412"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adjusted_rand_score(y_test, labels)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Try different number of clusters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster number and Score is:  2 0.27497894\n",
      "Cluster number and Score is:  3 0.27732375\n",
      "Cluster number and Score is:  4 0.27161542\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n",
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n",
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n",
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster number and Score is:  5 0.274734\n",
      "Cluster number and Score is:  6 0.29575256\n",
      "Cluster number and Score is:  7 0.28825805\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n",
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n",
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cluster number and Score is:  8 0.30210757\n",
      "Cluster number and Score is:  9 0.28578967\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/priya/miniconda3/envs/exptt/lib/python3.6/site-packages/sklearn/metrics/pairwise.py:259: RuntimeWarning: invalid value encountered in sqrt\n",
      "  return distances if squared else np.sqrt(distances, out=distances)\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import silhouette_score\n",
    "\n",
    "for cluster in range(2,10):\n",
    "    kmeans = KMeans(n_clusters=cluster, init='random', verbose =0).fit(intermediate_output)\n",
    "    labels = kmeans.labels_\n",
    "    score = silhouette_score(intermediate_output, labels, metric='euclidean')\n",
    "    print(\"Cluster number and Score is: \", cluster, score)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Visualizations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### 8 clusters\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "cluster = 8\n",
    "kmeans = KMeans(n_clusters=cluster, init='random', verbose =0).fit(intermediate_output)\n",
    "labels = kmeans.labels_\n",
    "\n",
    "mat = confusion_matrix(y_test, labels)\n",
    "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False,\n",
    "            xticklabels=dict_genres.keys())\n",
    "plt.xlabel('true label')\n",
    "plt.ylabel('predicted label');\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### 4 clusters\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "cluster = 4\n",
    "kmeans = KMeans(n_clusters=cluster, init='random', verbose =0).fit(intermediate_output)\n",
    "labels = kmeans.labels_\n",
    "\n",
    "mat = confusion_matrix(y_test, labels)\n",
    "sns.heatmap(mat.T, square=True, annot=True, fmt='d', cbar=False,\n",
    "            xticklabels=dict_genres.keys())\n",
    "plt.xlabel('true label')\n",
    "plt.ylabel('predicted label');\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:exptt]",
   "language": "python",
   "name": "conda-env-exptt-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}