Add naive bayes classifier notebook

Author: abdullahkady

naive-bayes.ipynb

@@ -0,0 +1,187 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "import numpy as np\n",
+    "import math"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_t(labels_path):\n",
+    "    with open(labels_path) as f:\n",
+    "        training_labels = [int(x) for x in f.read().splitlines()]\n",
+    "    return np.array(training_labels)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_confusion_matrix(true_labels, predictions):    \n",
+    "    conf_matrix = np.zeros((10, 10))\n",
+    "    for i, predicted_class in enumerate(predictions):\n",
+    "        conf_matrix[true_labels[i]][predicted_class] += 1\n",
+    "\n",
+    "    return conf_matrix"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def get_x(folder_name, number_images):\n",
+    "    x_input_points = np.zeros((0, 784))\n",
+    "    for i in range(1, number_images + 1): # +1 Since it's exclusive\n",
+    "        img_path = '{}/{}.jpg'.format(folder_name, i)\n",
+    "        x_input_points = np.append(x_input_points, plt.imread(img_path).reshape(1, 784), axis=0)\n",
+    "\n",
+    "    return x_input_points"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def gaussian(x, mean, variance):\n",
+    "    deno = (2 * math.pi * variance) ** 0.5\n",
+    "    exp = -1 * ( (x-mean)**2 / (2*variance) )\n",
+    "    return (1/deno) * math.exp(exp)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Loading in a separate cell to avoid multiple loads.\n",
+    "x_delta = get_x('Train', 2400)\n",
+    "training_true_labels = get_t('Train/Training Labels.txt')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "(10, 784)"
+      ]
+     },
+     "execution_count": 6,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "means = np.zeros((10, 784))\n",
+    "variances = np.zeros((10, 784))\n",
+    "\n",
+    "classes_inputs = np.split(x_delta/255, 10)\n",
+    "\n",
+    "for i, c in enumerate(classes_inputs):\n",
+    "    means[i] = np.mean(c, axis=0)\n",
+    "    \n",
+    "for i, c in enumerate(classes_inputs):\n",
+    "    variances[i] = np.var(c, axis=0)\n",
+    "    variances[i][variances[i] < 0.01] = 0.01\n",
+    "means.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stderr",
+     "output_type": "stream",
+     "text": [
+      "/home/abdullah/.virtualenvs/ml/lib/python3.5/site-packages/ipykernel_launcher.py:8: RuntimeWarning: overflow encountered in double_scalars\n",
+      "  \n"
+     ]
+    }
+   ],
+   "source": [
+    "x_delta_test = get_x('Test', 200)/255\n",
+    "test_true_labels = get_t('Test/Test Labels.txt')\n",
+    "\n",
+    "probabilities = np.ones((200, 10))\n",
+    "for i in range(200):\n",
+    "    for c in range(10):\n",
+    "        for f in range(784):\n",
+    "            probabilities[i][c] *= gaussian(x_delta_test[i][f], means[c][f], variances[c][f])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "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": [
+    "predictions = np.argmax(probabilities, axis=1)\n",
+    "conf_matrix = get_confusion_matrix(test_true_labels, predictions)\n",
+    "plt.imshow(conf_matrix)\n",
+    "plt.savefig('Confusion-naive.jpg')\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "file_extension": ".py",
+  "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": "3.5.2"
+  },
+  "mimetype": "text/x-python",
+  "name": "python",
+  "npconvert_exporter": "python",
+  "pygments_lexer": "ipython3",
+  "version": 3
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}