03_Assignment_1.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Introductory applied machine learning (INFR10069)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Assignment 1: Data analysis and visualisation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Marking Breakdown\n",
    "\n",
    "**70-100%** results/answer correct plus extra achievement at understanding or analysis of results. Clear explanations, evidence of creative or deeper thought will contribute to a higher grade.\n",
    "\n",
    "**60-69%** results/answer correct or nearly correct and well explained.\n",
    "\n",
    "**50-59%** results/answer in right direction but significant errors.\n",
    "\n",
    "**40-49%** some evidence that the student has gained some understanding, but not answered the questions\n",
    "properly.\n",
    "\n",
    "**0-39%** serious error or slack work."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Mechanics\n",
    "\n",
    "You should produce a Jupyter notebook in answer to this assignment.\n",
    "**You need to submit this notebook electronically as described below.**\n",
    "\n",
    "Place your notebook in a directory called `iamlans` and submit this directory using the submit command on a DICE machine. The format is:\n",
    "\n",
    "`submit iaml 1 iamlans`\n",
    "\n",
    "You can check the status of your submissions with the `show submissions` command.\n",
    "\n",
    "**Late submissions:** The policy stated in the School of Informatics MSc Degree Guide is that normally you will not be allowed to submit coursework late. See http://www.inf.ed.ac.uk/teaching/years/msc/courseguide10.html#exam for exceptions to this, e.g. in case of serious medical illness or serious personal problems.\n",
    "\n",
    "**Collaboration:** You may discuss the assignment with your colleagues, provided that the writing that you submit is entirely your own. That is, you should NOT borrow actual text or code from other students. We ask that you provide a list of the people who you've had discussions with (if any).\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Important Instructions\n",
    "\n",
    "1. In the following questions you are asked to run experiments using Python (version 2.7) and the following packages:\n",
    "    * Numpy\n",
    "    * Pandas\n",
    "    * Scikit-learn\n",
    "    * Matplotlib\n",
    "    * Seaborn\n",
    "\n",
    "2. Before you start make sure you have set up a vitual environment (or conda environment if you are working on your own machine) and the required packages installed. Instructions on how to set-up the working enviornment and install the required packages can be found in `01_Lab_1_Introduction`.\n",
    "\n",
    "3. Wherever you are required to produce code you should use code cells, otherwise you should use markdown cells to report results and explain answers.\n",
    "\n",
    "4. The .csv files that you will be using are located at `./datasets` (the `datasets` directory is adjacent to this file).\n",
    "\n",
    "5. **IMPORTANT:** Keep your answers brief and concise. Most questions can be answered with 2-3 lines of explanation (excluding coding questions)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports\n",
    "\n",
    "Execute the cell below to import all packages you will be using in the rest of the assignemnt."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from __future__ import print_function, division\n",
    "import os\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import sklearn as sklearn\n",
    "import seaborn as sns\n",
    "from sklearn.naive_bayes import MultinomialNB,GaussianNB\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Description of the dataset\n",
    "\n",
    "This assignment is based on the 20 Newsgroups Dataset. This dataset is a collection of approximately 20,000 newsgroup documents, partitioned (nearly) evenly across 20 different newsgroups, each corresponding to a different topic. Some of the newsgroups are very closely related to each other (e.g. comp.sys.ibm.pc.hardware, comp.sys.mac.hardware), while others are highly unrelated (e.g misc.forsale, soc.religion.christian). \n",
    "\n",
    "There are three versions of the 20 Newsgroups Dataset. In this assignment we will use the `bydate` matlab version in which documents are sorted by date into training (60%) and test (40%) sets, newsgroup-identifying headers are dropped and duplicates are removed. This collection comprises roughly 61,000 different words, which results in a bag-of-words representation with frequency counts. More specifically, each document is represented by a 61,000 dimensional vector that contains the counts for each of the 61,000 different words present in the respective document. \n",
    "\n",
    "To save you time and to make the problem manageable with limited computational resources, we preprocessed the original dataset. We will use documents from only 5 out of the 20 newsgroups, which results in a 5-class problem. More specifically the 5 classes correspond to the following newsgroups: \n",
    "1. `alt.atheism`\n",
    "2. `comp.sys.ibm.pc.hardware`\n",
    "3. `comp.sys.mac.hardware`\n",
    "4. `rec.sport.baseball`\n",
    "5. `rec.sport.hockey `\n",
    "\n",
    "However, note here that classes 2-3 and 4-5 are rather closely related. Additionally, we computed the [mutual information](https://en.wikipedia.org/wiki/Mutual_information) of each word with the class attribute and selected the 520 words out of 61,000 that had highest mutual information. Therefore, our dataset is a $N \\times 520$ dimensional matrix, where $N$ is the number of documents. For very sophisticated technical reasons 1 was added to all the word counts in part A. The resulting representation is much more compact and can be used directly to perform our experiments in Python."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Exploration of the dataset [40%]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Your first task is to get a feel for the data that you will be dealing with in the rest of the assignment."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.1 ==========\n",
    "Load the datasets `train_20news_partA.csv` and `train_20news_partB.csv` into two separate pandas DataFrames."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Your code goes here\n",
    "# Read the datasets csv, which is in 'datasets' folder of current working directory (iaml), and put it in separate dataframes\n",
    "part_a = pd.read_csv(os.path.join(os.getcwd(),'datasets','train_20news_partA.csv'),delimiter = ',')\n",
    "part_b = pd.read_csv(os.path.join(os.getcwd(),'datasets','train_20news_partB.csv'),delimiter = ',')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### ========== Question 1.2 ==========\n",
    "Display basic information for dataset A such as number of columns, type, and memory usage (*hint: pandas dataframes have a built in method for this*) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 2129 entries, 0 to 2256\n",
      "Columns: 521 entries, w1_aaa to class\n",
      "dtypes: int64(521)\n",
      "memory usage: 8.5 MB\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Your code goes here\n",
    "print (part_a.info()) # Panda's built-in method to display basic information of a dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.3 ==========\n",
    "How many data points and how many attributes are there in the dataset that we can use to model the target variable `class`?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "There are 2129 data points and 520 attributes in the dataset.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### ========== Question 1.4  ==========\n",
    "Use a Pandas method to display the summary statistics for the `news_A` DataFrame."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>w1_aaa</th>\n",
       "      <th>w2_pins</th>\n",
       "      <th>w3_kmr</th>\n",
       "      <th>w4_notion</th>\n",
       "      <th>w5_queens</th>\n",
       "      <th>w6_dwyer</th>\n",
       "      <th>w7_defenseman</th>\n",
       "      <th>w8_gld</th>\n",
       "      <th>w9_tocchet</th>\n",
       "      <th>w10_home</th>\n",
       "      <th>...</th>\n",
       "      <th>w512_constantly</th>\n",
       "      <th>w513_generate</th>\n",
       "      <th>w514_definite</th>\n",
       "      <th>w515_lacks</th>\n",
       "      <th>w516_combination</th>\n",
       "      <th>w517_sitting</th>\n",
       "      <th>w518_surface</th>\n",
       "      <th>w519_fashion</th>\n",
       "      <th>w520_sit</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.00000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "      <td>2129.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>5.128229</td>\n",
       "      <td>6.097698</td>\n",
       "      <td>6.120244</td>\n",
       "      <td>5.551902</td>\n",
       "      <td>5.521841</td>\n",
       "      <td>6.12776</td>\n",
       "      <td>5.633161</td>\n",
       "      <td>6.090653</td>\n",
       "      <td>5.970409</td>\n",
       "      <td>5.624706</td>\n",
       "      <td>...</td>\n",
       "      <td>9.666510</td>\n",
       "      <td>9.217473</td>\n",
       "      <td>9.061531</td>\n",
       "      <td>9.398309</td>\n",
       "      <td>9.175200</td>\n",
       "      <td>9.708783</td>\n",
       "      <td>8.807891</td>\n",
       "      <td>9.719587</td>\n",
       "      <td>9.307656</td>\n",
       "      <td>3.092532</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>38.292577</td>\n",
       "      <td>46.190953</td>\n",
       "      <td>46.656022</td>\n",
       "      <td>40.953913</td>\n",
       "      <td>40.978098</td>\n",
       "      <td>45.96434</td>\n",
       "      <td>41.146918</td>\n",
       "      <td>45.762060</td>\n",
       "      <td>44.266628</td>\n",
       "      <td>40.769105</td>\n",
       "      <td>...</td>\n",
       "      <td>45.844064</td>\n",
       "      <td>43.948910</td>\n",
       "      <td>40.969185</td>\n",
       "      <td>43.833064</td>\n",
       "      <td>42.403283</td>\n",
       "      <td>47.294120</td>\n",
       "      <td>39.341038</td>\n",
       "      <td>46.185082</td>\n",
       "      <td>45.059367</td>\n",
       "      <td>1.395948</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>2.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>3.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.00000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>4.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>572.000000</td>\n",
       "      <td>583.000000</td>\n",
       "      <td>579.000000</td>\n",
       "      <td>580.000000</td>\n",
       "      <td>591.000000</td>\n",
       "      <td>600.00000</td>\n",
       "      <td>546.000000</td>\n",
       "      <td>591.000000</td>\n",
       "      <td>578.000000</td>\n",
       "      <td>578.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>590.000000</td>\n",
       "      <td>587.000000</td>\n",
       "      <td>577.000000</td>\n",
       "      <td>598.000000</td>\n",
       "      <td>568.000000</td>\n",
       "      <td>599.000000</td>\n",
       "      <td>585.000000</td>\n",
       "      <td>600.000000</td>\n",
       "      <td>597.000000</td>\n",
       "      <td>5.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 521 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            w1_aaa      w2_pins       w3_kmr    w4_notion    w5_queens  \\\n",
       "count  2129.000000  2129.000000  2129.000000  2129.000000  2129.000000   \n",
       "mean      5.128229     6.097698     6.120244     5.551902     5.521841   \n",
       "std      38.292577    46.190953    46.656022    40.953913    40.978098   \n",
       "min       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
       "25%       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
       "50%       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
       "75%       1.000000     1.000000     1.000000     1.000000     1.000000   \n",
       "max     572.000000   583.000000   579.000000   580.000000   591.000000   \n",
       "\n",
       "         w6_dwyer  w7_defenseman       w8_gld   w9_tocchet     w10_home  \\\n",
       "count  2129.00000    2129.000000  2129.000000  2129.000000  2129.000000   \n",
       "mean      6.12776       5.633161     6.090653     5.970409     5.624706   \n",
       "std      45.96434      41.146918    45.762060    44.266628    40.769105   \n",
       "min       1.00000       1.000000     1.000000     1.000000     1.000000   \n",
       "25%       1.00000       1.000000     1.000000     1.000000     1.000000   \n",
       "50%       1.00000       1.000000     1.000000     1.000000     1.000000   \n",
       "75%       1.00000       1.000000     1.000000     1.000000     1.000000   \n",
       "max     600.00000     546.000000   591.000000   578.000000   578.000000   \n",
       "\n",
       "          ...       w512_constantly  w513_generate  w514_definite  \\\n",
       "count     ...           2129.000000    2129.000000    2129.000000   \n",
       "mean      ...              9.666510       9.217473       9.061531   \n",
       "std       ...             45.844064      43.948910      40.969185   \n",
       "min       ...              1.000000       1.000000       1.000000   \n",
       "25%       ...              3.000000       2.000000       3.000000   \n",
       "50%       ...              5.000000       5.000000       5.000000   \n",
       "75%       ...              7.000000       7.000000       7.000000   \n",
       "max       ...            590.000000     587.000000     577.000000   \n",
       "\n",
       "        w515_lacks  w516_combination  w517_sitting  w518_surface  \\\n",
       "count  2129.000000       2129.000000   2129.000000   2129.000000   \n",
       "mean      9.398309          9.175200      9.708783      8.807891   \n",
       "std      43.833064         42.403283     47.294120     39.341038   \n",
       "min       1.000000          1.000000      1.000000      1.000000   \n",
       "25%       3.000000          2.000000      3.000000      3.000000   \n",
       "50%       5.000000          5.000000      5.000000      5.000000   \n",
       "75%       7.000000          7.000000      7.000000      7.000000   \n",
       "max     598.000000        568.000000    599.000000    585.000000   \n",
       "\n",
       "       w519_fashion     w520_sit        class  \n",
       "count   2129.000000  2129.000000  2129.000000  \n",
       "mean       9.719587     9.307656     3.092532  \n",
       "std       46.185082    45.059367     1.395948  \n",
       "min        1.000000     1.000000     1.000000  \n",
       "25%        3.000000     2.000000     2.000000  \n",
       "50%        5.000000     4.000000     3.000000  \n",
       "75%        7.000000     6.000000     4.000000  \n",
       "max      600.000000   597.000000     5.000000  \n",
       "\n",
       "[8 rows x 521 columns]"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "part_a.describe() # Panda's built-in method to display the summary statistics of a dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### ========== Question 1.5 ==========\n",
    "Display the first 7 instances of dataset A."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>w1_aaa</th>\n",
       "      <th>w2_pins</th>\n",
       "      <th>w3_kmr</th>\n",
       "      <th>w4_notion</th>\n",
       "      <th>w5_queens</th>\n",
       "      <th>w6_dwyer</th>\n",
       "      <th>w7_defenseman</th>\n",
       "      <th>w8_gld</th>\n",
       "      <th>w9_tocchet</th>\n",
       "      <th>w10_home</th>\n",
       "      <th>...</th>\n",
       "      <th>w512_constantly</th>\n",
       "      <th>w513_generate</th>\n",
       "      <th>w514_definite</th>\n",
       "      <th>w515_lacks</th>\n",
       "      <th>w516_combination</th>\n",
       "      <th>w517_sitting</th>\n",
       "      <th>w518_surface</th>\n",
       "      <th>w519_fashion</th>\n",
       "      <th>w520_sit</th>\n",
       "      <th>class</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>6</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>...</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>6</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>7</td>\n",
       "      <td>8</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>6</td>\n",
       "      <td>1</td>\n",
       "      <td>7</td>\n",
       "      <td>8</td>\n",
       "      <td>6</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>8</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>7</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "      <td>7</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>...</td>\n",
       "      <td>3</td>\n",
       "      <td>6</td>\n",
       "      <td>8</td>\n",
       "      <td>8</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>7 rows × 521 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   w1_aaa  w2_pins  w3_kmr  w4_notion  w5_queens  w6_dwyer  w7_defenseman  \\\n",
       "0       1        1       1          1          1         1              1   \n",
       "1       1        1       1          1          1         1              1   \n",
       "2       1        1       1          1          1         1              1   \n",
       "3       1        1       1          1          1         1              1   \n",
       "4       1        1       1          1          1         1              1   \n",
       "5       1        1       1          1          1         1              1   \n",
       "6       1        1       1          1          2         1              1   \n",
       "\n",
       "   w8_gld  w9_tocchet  w10_home  ...    w512_constantly  w513_generate  \\\n",
       "0       1           1         1  ...                  1              5   \n",
       "1       1           1         1  ...                  2              3   \n",
       "2       1           1         2  ...                  4              2   \n",
       "3       1           1         1  ...                  6              3   \n",
       "4       1           1         1  ...                  6              1   \n",
       "5       1           1         1  ...                  7              5   \n",
       "6       1           1         1  ...                  3              6   \n",
       "\n",
       "   w514_definite  w515_lacks  w516_combination  w517_sitting  w518_surface  \\\n",
       "0              3           6                 8             4             6   \n",
       "1              6           8                 3             4             5   \n",
       "2              6           5                 2             5             5   \n",
       "3              6           1                 1             8             1   \n",
       "4              7           8                 6             3             1   \n",
       "5              4           5                 5             7             7   \n",
       "6              8           8                 3             5             2   \n",
       "\n",
       "   w519_fashion  w520_sit  class  \n",
       "0             4         8      4  \n",
       "1             6         5      4  \n",
       "2             7         8      2  \n",
       "3             8         4      3  \n",
       "4             8         4      3  \n",
       "5             7         2      2  \n",
       "6             1         2      1  \n",
       "\n",
       "[7 rows x 521 columns]"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "part_a.head(7)  # Panda's built-in method to take n first entries of a dataframe"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.6 ==========\n",
    "Display the names of the first 100 attributes in dataset A. \n",
    "\n",
    "You might observe that each attribute consists of two parts:\n",
    "1. `w<x>_` (where x is an index corresponding to each word)\n",
    "2. the actual name of the word"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "w1_aaa\n",
      "w2_pins\n",
      "w3_kmr\n",
      "w4_notion\n",
      "w5_queens\n",
      "w6_dwyer\n",
      "w7_defenseman\n",
      "w8_gld\n",
      "w9_tocchet\n",
      "w10_home\n",
      "w11_buying\n",
      "w12_internet\n",
      "w13_slots\n",
      "w14_compatible\n",
      "w15_transfer\n",
      "w16_baltimore\n",
      "w17_mean\n",
      "w18_person\n",
      "w19_performance\n",
      "w20_support\n",
      "w21_tor\n",
      "w22_gm\n",
      "w23_mouse\n",
      "w24_base\n",
      "w25_population\n",
      "w26_bob\n",
      "w27_set\n",
      "w28_it\n",
      "w29_earth\n",
      "w30_faith\n",
      "w31_steve\n",
      "w32_caps\n",
      "w33_printer\n",
      "w34_east\n",
      "w35_cable\n",
      "w36_adapter\n",
      "w37_mss\n",
      "w38_catcher\n",
      "w39_bullpen\n",
      "w40_obp\n",
      "w41_innocent\n",
      "w42_european\n",
      "w43_angeles\n",
      "w44_settings\n",
      "w45_words\n",
      "w46_rit\n",
      "w47_shots\n",
      "w48_ports\n",
      "w49_vga\n",
      "w50_coverage\n",
      "w51_jumpers\n",
      "w52_bases\n",
      "w53_sea\n",
      "w54_pts\n",
      "w55_behavior\n",
      "w56_domi\n",
      "w57_sabres\n",
      "w58_yzerman\n",
      "w59_messier\n",
      "w60_goalies\n",
      "w61_hawks\n",
      "w62_our\n",
      "w63_sx\n",
      "w64_view\n",
      "w65_hitters\n",
      "w66_richard\n",
      "w67_point\n",
      "w68_nyi\n",
      "w69_mvp\n",
      "w70_kill\n",
      "w71_nl\n",
      "w72_field\n",
      "w73_connector\n",
      "w74_stars\n",
      "w75_th\n",
      "w76_install\n",
      "w77_traded\n",
      "w78_configuration\n",
      "w79_standard\n",
      "w80_rotation\n",
      "w81_ultb\n",
      "w82_sports\n",
      "w83_pds\n",
      "w84_canada\n",
      "w85_did\n",
      "w86_clock\n",
      "w87_first\n",
      "w88_switch\n",
      "w89_tonight\n",
      "w90_record\n",
      "w91_singer\n",
      "w92_majors\n",
      "w93_royals\n",
      "w94_does\n",
      "w95_flames\n",
      "w96_of\n",
      "w97_series\n",
      "w98_plays\n",
      "w99_det\n",
      "w100_pitched\n"
     ]
    }
   ],
   "source": [
    "# Your code goes here\n",
    "attribute_name = list(part_a.columns.values) # Put all of the attribute name of the dataset into a list\n",
    "for i in range(0,100):\n",
    "    print(attribute_name[i]) # Print the first 100 attributes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.7 ==========\n",
    "Familiarise yourself with the [`stripplot`](https://stanford.edu/~mwaskom/software/seaborn/generated/seaborn.stripplot.html) function in `seaborn`. Pick one attribute of your choice (except `class`) and display a stripplot for that attribute for dataset A. Demonstrate the distribution of the data separately for each class (by making appropriate use of the `x` argument in `stripplot`). Set the `jitter` argument to `True` and the `alpha` argument to an appropriate value (to add transparency). When the jitter parameter is enabled a small amount of noise is added to the data so that there is less overlap and the distribution is easier to visualise."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x3d23240>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXEAAAEMCAYAAAAyO4a7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n4iIiDlOIi4g4TCEuIuKw/w+GWM1/KLm/wAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x3d232e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "sns.stripplot(x=part_a.iloc[:,65],jitter=.2,alpha=.5) # Display the distribution of occurence for word in index 65 (w66_richard)in emails\n",
    "\n",
    "# There are some outliers, in which word 'richard' appeared more than 100 times in a single email"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.8 ==========\n",
    "The stripplot illustrates the distribution of a single attribute. We can also visualise the joint distribution of two variables by using a scatter plot. Again, we want to add a bit of noise into the data so that is easier to see which parts of the space (2-dimensional in our case) have larger probability densities. \n",
    "\n",
    "For this, you will be using the function `scatter_jitter` provided below. This function takes as input two numpy arrays containing the features of interest. Pick two attributes of your choice from dataset A and use the provided function to plot their joint distribution. You can play around with the amount of noise added by tweaking the `jitter` parameter. Alternatively, you can just use its default value which is set to 0.2."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def scatter_jitter(arr1, arr2, jitter=0.2):\n",
    "    \"\"\" Plots a joint scatter plot of two arrays by adding small noise to each example. \n",
    "    Noise is proportional to variance in each dimension. \"\"\"\n",
    "    arr1 = np.asarray(arr1)\n",
    "    arr2 = np.asarray(arr2)\n",
    "    arr1 = arr1 + jitter*arr1.std(axis=0)*np.random.standard_normal(arr1.shape)\n",
    "    arr2 = arr2 + jitter*arr2.std(axis=0)*np.random.standard_normal(arr2.shape)\n",
    "    plt.scatter(arr1, arr2, marker=4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAECCAYAAAAciLtvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlAxERAQlAxERQclARERQMhAREeD/A5Q5Xj2QoCd/AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xd914780>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "scatter_jitter(part_a['w51_jumpers'],part_a['w66_richard'],jitter=0.5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.9 ==========\n",
    "From the strip and scatter plots above you might observe that there is something peculiar about the data. Indeed most attributes take very small values (usually in the range 1-10) but there are some data points (i.e. rows) in the dataset where the attributes take very large values. These data points are called [outliers](https://en.wikipedia.org/wiki/Outlier).\n",
    "\n",
    "You might think that the presence of outliers in the dataset has been a resut of noise contamination (you wouldn't expect the same word to appear 600 times within an e-mail, would you?). Your job now is to create a new dataset from dataset A (name it `news_A_clean`) and remove the outliers. Create some metric to find the outliers and check that your metric is reasonable. Be careful not to alter the original `news_A`...we may require it in its dirty format later..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2129, 521)\n",
      "(2099, 521)\n"
     ]
    }
   ],
   "source": [
    "# Your code goes here\n",
    "part_a_copy = part_a.copy() #copy into another dataframe\n",
    "attribute_name = list(part_a_copy.columns.values) #get column names\n",
    "\n",
    "# One way to clean the dataset, is to drop a row if it contains a certain value\n",
    "# or in our case, if an email (an observation, a row) contains a word (an attributes,a column) which appear > 100 times \n",
    "\n",
    "#news_A_clean = part_a_copy(part_a_copy<100) #create new df that change the value that > 100 to NaN\n",
    "#news_A_clean.dropna() #drop the row that contains NaN value in one of its column\n",
    "\n",
    "# Assuming the dataset has normal distribution,\n",
    "# dataset can also be cleaned using z-score (standard score), which indicates how many standard deviations an element is from the mean. \n",
    "for i in range (len(attribute_name)-1): #for each column, except the last one which is our output\n",
    "    attribute_mean = part_a_copy[attribute_name[i]].mean() #mean value of each attribute\n",
    "    attribute_std = part_a_copy[attribute_name[i]].std() #standard deviation value of each attribute\n",
    "    news_A_clean = part_a_copy[np.abs((part_a_copy[attribute_name[i]]-attribute_mean)/attribute_std)<=2]\n",
    "    #keep only row which has a z_score <= 2, which then will give us around 95% of the data\n",
    "print (part_a_copy.shape)\n",
    "print (news_A_clean.shape)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 1.10 ==========\n",
    "Write code to return the number of data points in the clean dataset, and the number of documents that have been excluded as outliers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2099, 521)\n",
      "2099 datapoints\n",
      "30 documents\n"
     ]
    }
   ],
   "source": [
    "# Your code goes here\n",
    "print(news_A_clean.shape)\n",
    "dp = len(news_A_clean.index) #number of data points in the clean dataset, number of rows x number of columns \n",
    "print (dp,\"datapoints\")\n",
    "de = len(part_a_copy.index) -  len(news_A_clean.index) #number of documents that have been excluded\n",
    "print (de,\"documents\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Naive Bayes classification [60%]\n",
    "Now we want to fit a Gaussian Naive Bayes model to the cleaned dataset A. You might want first to familiarise yourself with the [`GaussianNB`](http://scikit-learn.org/stable/modules/generated/sklearn.naive_bayes.GaussianNB.html) class in `Sklearn`."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.1 ==========\n",
    "\n",
    "By using the `scatter_jitter` function provided above, display a scatter plot of the features `w281_ico` and `w273_tek` for the cleaned dataset A. Set the jitter value to something small (e.g. 0.1). Label axes appropriately."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.text.Text at 0xf98c5c0>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXsAAAERCAYAAACaUQc3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xf0c67b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "scatter_jitter(news_A_clean['w281_ico'],news_A_clean['w273_tek'],jitter=0.1)\n",
    "plt.xlabel('w281_ico')\n",
    "plt.ylabel('w273_tek')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.2 ==========\n",
    "What do you observe? \n",
    "\n",
    "How does that relate to the Naive Bayes assumption? \n",
    "\n",
    "What would be the main issue we would have to face if we didn't make this assumption?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Your answer goes here:*\n",
    "\n",
    "By the scatter plot, it can be seen that 'w281_ico'  and 'w273_tek' have positive correlation, where both the data move in common into more positive value.\n",
    "\n",
    "Naive Bayes assumes that all attributes have no correlation to each other, or in other words, each attributes dependants to any other attribute.\n",
    "\n",
    "But since we have a lot of attributes, if we didn't make the assumption, probability of one attribute given another attributes for each attributes we have should be calculated, which is a lot of computation."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.3 ==========\n",
    "Fit a Gaussian Naive Bayes model to the cleaned dataset A. Your input features should be all the attributes in the dataset except the `class` attribute which will be your target. Display the classification accuracy on the training dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification accuracy: 0.878037160553\n"
     ]
    }
   ],
   "source": [
    "# Your code goes here\n",
    "x = news_A_clean.drop('class',axis=1) #drop the class attribute, axis 1 denotes that we refer to column not row\n",
    "y = pd.Series(news_A_clean['class']) #create another series for the class\n",
    "\n",
    "clf_gauss = GaussianNB() #initiate the classifier\n",
    "\n",
    "clf_gauss.fit(x,y) #fit with the data\n",
    "\n",
    "print ('Classification accuracy:',clf_gauss.score(x,y,sample_weight=None)) #classification accuracy score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.4 ==========\n",
    "Plot the (normalised) confusion matrix for the training data. Label axes appropriately."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 0.96694215  0.0023753   0.02606635  0.          0.        ]\n",
      " [ 0.          0.64608076  0.35308057  0.          0.        ]\n",
      " [ 0.          0.01900238  0.98104265  0.          0.        ]\n",
      " [ 0.00275482  0.00475059  0.06635071  0.93064877  0.        ]\n",
      " [ 0.          0.0023753   0.03317536  0.0917226   0.87443946]]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAEZCAYAAAC+bm+MAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwuQkUQshhMn9PyyMhyRj52fTAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x124fdac8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "cm = confusion_matrix(y, clf_gauss.predict(x))\n",
    "cm_norm = cm /cm.sum(axis=1) #axis 1 means the actual (horizontal), 0 is the predicted (vertical)\n",
    "print (cm_norm)\n",
    "\n",
    "def plot_confusion_matrix(cm, classes=None, title='Confusion matrix'):\n",
    "    \"\"\"Plots a confusion matrix.\"\"\"\n",
    "    if classes is not None:\n",
    "        sns.heatmap(cm, xticklabels=classes, yticklabels=classes, vmin=0., vmax=1., annot=True)\n",
    "    else:\n",
    "        sns.heatmap(cm)\n",
    "    plt.title(title)\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    \n",
    "plot_confusion_matrix(cm_norm, classes=[1,2,3,4,5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.5  ==========\n",
    "\n",
    "Comment on the confusion matrix from the previous question. Does it look like what you would have expected? Explain."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Your answer goes here*\n",
    "It looks just like what would have been expected, as the confusion matrix shows a big value across the diagonal, where the prediction meets the true value.\n",
    "However, there is one particular class, class 1, on which only 65% of the prediction that meets the target value."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.6 ==========\n",
    "Fit a Gaussian Naive Bayes model to the original dataset A (including the outliers). Display the classification accuracy on the training dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification accuracy: 0.205260685768\n"
     ]
    }
   ],
   "source": [
    "# Your code goes here\n",
    "x_ori = part_a.drop('class',axis=1) # Drop 'class' column. Axis 1 denotes that we refer to column not row\n",
    "y_ori = pd.Series(part_a['class']) # Create another series for that 'class'\n",
    "\n",
    "clf_gauss_ori = GaussianNB() # Initiate the Gaussian Naive Bayes classifier\n",
    "\n",
    "clf_gauss_ori.fit(x_ori,y_ori) # fit it with the data\n",
    "\n",
    "print ('Classification accuracy:',clf_gauss_ori.score(x_ori,y_ori,sample_weight=None)) # Classification accuracy score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.7 ==========\n",
    "Comment on the above results (Questions 2.3 & 2.6). In particular explain why you think that cleaning the data helps in this case."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Your answer goes here*\n",
    "\n",
    "Because outliers in the dataset would make our classifier too focused to also fit that outlier, which then make our model less general. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.8 ==========\n",
    "\n",
    "Now we want to evaluate the generalisation of the classifier on new (i.e. unseen data). Use the classifier you trained in Question 2.5 (i.e. on the cleaned dataset) and test its performance on dataset `train_20news_partB`. \n",
    "\n",
    "Display the (normalized) confusion matrix and the classification accuracy on the Dataset B."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "128\n",
      "Classification accuracy: 0.8203125\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAEZCAYAAAC+bm+MAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nivIlAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0xf97b390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "x_test = part_b.drop('class',axis=1)\n",
    "y_test= pd.Series(part_b['class']) \n",
    "print (len(x_test))\n",
    "cm_test = confusion_matrix(y_test, clf_gauss.predict(x_test))\n",
    "cm_norm_test = cm_test /cm_test.sum(axis=1)\n",
    "    \n",
    "plot_confusion_matrix(cm_norm_test,classes=[1,2,3,4,5])\n",
    "print ('Classification accuracy:',clf_gauss.score(x_test,y_test,sample_weight=None)) #classification accuracy score\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.9 ==========\n",
    "\n",
    "Comment on the results from the previous question. Do you think this is an acceptable level of performance? Which are the easiest and most difficult classes to predict correctly? "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Your answer goes here*\n",
    "Yes, because most of our prediction fit the true value (all value in the confusion matrix diagonal have a rather big value. \n",
    "\n",
    "The easiest class to predict is class 1, which our classifier predict 95% of the data in class 1 correctly.\n",
    "The hardest class to predict must be class 2, which only 59% predicted correctly."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.10 ==========\n",
    "What is a reasonable baseline against which to compare the classiffication performance? *Hint: What is the simplest classiffier you can think of and what would its performance be on this dataset?* "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Your answer goes here*\n",
    "A random classifier, which will give a random class for any given input.\n",
    "The performance on this dataset would be quite low."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.11 ==========\n",
    "\n",
    "Estimate the baseline performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification accuracy: 0.169843906094\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAEZCAYAAAC+bm+MAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nY3DRUHFFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12685630>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Your code goes here\n",
    "xran_test = part_b.drop('class',axis=1)\n",
    "yran_test = pd.Series(part_b['class']) \n",
    "\n",
    "#for every input, the classifier will generate any random value, in range 0 - 5 (values of 'class')\n",
    "yran_predict = np.random.randint(5,size=len(xran_test))+1 \n",
    "\n",
    "cmran_test = confusion_matrix(yran_test, yran_predict)\n",
    "cmran_norm_test = cmran_test /cmran_test.sum(axis=1)\n",
    "    \n",
    "plot_confusion_matrix(cmran_norm_test,classes=[1,2,3,4,5])\n",
    "\n",
    "print ('Classification accuracy:',(np.trace(cmran_norm_test)/5)) #classification accuracy score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The baseline performance estimations depicts in the confusion matrix above, where the predicted and true data do not meet much."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.12 ==========\n",
    "\n",
    "Execute the cell below to get the prediction on the test dataset by using a different classifier which we will be introducing in this class later on. By using this prediction provided below (`rf_prediction`) plot the confusion matrix and display the classification accuracy on the test dataset. *Important: Make sure the test dataset is loaded in a DataFrame called `news_B` otherwise execution will return an error. In that case replace the DataFrame name in the third line.*"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification accuracy on the test set by using a Random Forest: 0.8671875\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWoAAAEZCAYAAAC+bm+MAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11f18cf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "#to fit all the required variable names\n",
    "X_tr = x.copy() #copy of the cleaned news A dataset (except column 'class') to X_tr\n",
    "y_tr = y.copy() #copy of the cleaned news A column 'class' to y_tr\n",
    "news_B = pd.read_csv(part_b_path,delimiter = ',') #import the test dataset into a new variable called news_B\n",
    "\n",
    "rf = RandomForestClassifier(n_estimators = 50).fit(X=X_tr, y=y_tr)\n",
    "X_ts = news_B.drop('class', axis=1)\n",
    "y_ts = pd.Series(news_B['class']) #define y_ts, column 'class' of news_B\n",
    "rf_prediction = rf.predict(X=X_ts)\n",
    "# Your code goes here\n",
    "print('Classification accuracy on the test set by using a Random Forest:', accuracy_score(y_ts, rf.predict(X=X_ts)))\n",
    "plt.figure()\n",
    "classes=['0','1','2','3','4']\n",
    "cm = confusion_matrix(y_ts, rf_prediction)\n",
    "cm_norm = cm/cm.sum(axis=1)[:, np.newaxis]\n",
    "plot_confusion_matrix(cm_norm, classes=classes)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ========== Question 2.13 ==========\n",
    "\n",
    "Which classifier (Naive Bayes or Random Forest) would you trust if you had to choose? What are the reasons you believe the Gaussian Naive Bayes classifier does not perofm so well in this particular problem? You are not expected to justify the performance level achieved by the Random Forest classifier."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "*Your answer goes here.*\n",
    "Random Forest would be more trustworthy. In this particular problem, we have already found that at least there are 2 attributes (w281_ico & w273_tek) that correlate with each other and thus we should not assume that all attributes are independent to each other. "
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [iaml_env]",
   "language": "python",
   "name": "Python [iaml_env]"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}