Digit_Recognition.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MNIST dataset classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Author ->** Simranjeet Singh - Data Scientist - GNA University"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import scipy as sp\n",
    "\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import StratifiedShuffleSplit\n",
    "from sklearn.externals import joblib\n",
    "\n",
    "np.random.seed(42)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Fetching the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100% [................................................................................] 4542 / 4542"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'t10k-labels-idx1-ubyte.gz'"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import wget as w\n",
    "w.download('http://yann.lecun.com/exdb/mnist/train-images-idx3-ubyte.gz')\n",
    "w.download('http://yann.lecun.com/exdb/mnist/train-labels-idx1-ubyte.gz')\n",
    "w.download('http://yann.lecun.com/exdb/mnist/t10k-images-idx3-ubyte.gz')\n",
    "w.download('http://yann.lecun.com/exdb/mnist/t10k-labels-idx1-ubyte.gz')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Loading the data to Numpy arrays"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"train-images.idx3-ubyte\", \"rb\") as f:\n",
    "    X = np.frombuffer(f.read(), dtype=np.uint8, offset=16).copy()\n",
    "    X = X.reshape((60000, 28*28))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"train-labels.idx1-ubyte\", \"rb\") as f:\n",
    "    y = np.frombuffer(f.read(), dtype=np.uint8, offset=8)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing 64 random images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 504x504 with 64 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ind = np.random.randint(0, 60000, 64)\n",
    "X_vis = X[ind, :].reshape(64, 28, 28) \n",
    "y_vis = y[ind]\n",
    "plt.figure(figsize=(7, 7))\n",
    "for i in range(1,9):\n",
    "    for j in range(1, 9):\n",
    "        plt.subplot(8, 8, (i-1)*8+j)\n",
    "        plt.imshow(X_vis[(i-1)*8+j-1], cmap=plt.cm.Greys)\n",
    "        plt.gca().xaxis.set_ticks([])\n",
    "        plt.gca().yaxis.set_ticks([])\n",
    "        plt.title(y_vis[(i-1)*8+j-1])\n",
    "        \n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Splitting the dataset. Making training and validation sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "sss = StratifiedShuffleSplit(n_splits=1, test_size=0.3)\n",
    "indx = sss.split(X, y)\n",
    "\n",
    "for train_index, val_index in indx:\n",
    "    X_train, X_val = X[train_index], X[val_index]\n",
    "    y_train, y_val = y[train_index], y[val_index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training data:  (42000, 784) (42000,)\n",
      "Test data:  (42000, 784) (42000,)\n"
     ]
    }
   ],
   "source": [
    "print('Training data: ', X_train.shape, y_train.shape)\n",
    "print('Test data: ', X_train.shape, y_train.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy is 0.9430555555555555 %\n"
     ]
    }
   ],
   "source": [
    "clf = RandomForestClassifier(n_estimators=10, criterion='entropy', n_jobs=-1, random_state=42)\n",
    "clf.fit(X_train, y_train)\n",
    "print('Accuracy is', sum(y_val == clf.predict(X_val))/1./y_val.shape[0], '%')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "scores = []\n",
    "trees =  [5, 10, 20, 30, 40, 50, 60, 100, 300]\n",
    "for n_trees in trees:\n",
    "    clf = RandomForestClassifier(n_estimators=n_trees, n_jobs=-1, random_state=42)\n",
    "    clf.fit(X_train, y_train)\n",
    "    scores.append(100*sum(y_val == clf.predict(X_val))/1./y_val.shape[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure()\n",
    "plt.plot(trees, scores)\n",
    "plt.plot([trees[np.argmax(scores)]], [max(scores)], 'ro')\n",
    "plt.xlabel('# trees')\n",
    "plt.ylabel('Accuracy (%)')\n",
    "plt.xlim(min(trees)-10, max(trees)+10)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Best number of trees: 300\n",
      "Best score:  96.77777777777777\n"
     ]
    }
   ],
   "source": [
    "print('Number of Trees:', trees[np.argmax(scores)])\n",
    "print('Score: ', max(scores))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Retraining the classifier on the whole dataset and testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=300, n_jobs=-1,\n",
       "            oob_score=False, random_state=42, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=42)\n",
    "clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Getting the MNIST test data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"t10k-images.idx3-ubyte\", \"rb\") as f:\n",
    "    X_test = np.frombuffer(f.read(), dtype=np.uint8, offset=16).copy()  \n",
    "    X_test = X_test.reshape((10000, 28*28))\n",
    "    \n",
    "with open(\"t10k-labels.idx1-ubyte\", \"rb\") as f:\n",
    "    y_test = np.frombuffer(f.read(), dtype=np.uint8, offset=8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy on the TEST set is 97.15 %\n"
     ]
    }
   ],
   "source": [
    "print('Accuracy is', 100*sum(y_test == clf.predict(X_test))/1./y_test.shape[0], '%')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us now retrain the classifier on the whole dataset in order to learn more variability in the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "            max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "            min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "            min_samples_leaf=1, min_samples_split=2,\n",
       "            min_weight_fraction_leaf=0.0, n_estimators=300, n_jobs=-1,\n",
       "            oob_score=False, random_state=42, verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = RandomForestClassifier(n_estimators=300, n_jobs=-1, random_state=42)\n",
    "clf.fit(np.vstack((X, X_test)), np.hstack((y,y_test)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## \"Pickling\" the trained classifier\n",
    "\n",
    "Pickle mechanism has some security issues, so, carefully read the instructions on the scikit-learn web-site"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['digits_classifier/clf.pkl']"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "joblib.dump(clf, 'Digit_App/clf.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": 1
}