ex6_spam.py

""" Machine Learning Online Class
Exercise 6 | Spam Classification with SVMs
Instructions
------------
This file contains code that helps you get started on the
exercise. You will need to complete the following functions:
   gaussianKernel
   dataset3Params
   processEmail
   emailFeatures
For this exercise, you will not need to change any code in this file,
or any other files other than those mentioned above.
"""

# Imports
import ex6_spamHelper as spamHelper
import numpy as np
import scipy.io as io
from sklearn import svm
from data import vocab


def main():
    #  ==================== Part 1: Email Preprocessing ====================
    #  To use an SVM to classify emails into Spam v.s. Non-Spam, you first need
    #  to convert each email into a vector of features. In this part, you will
    #  implement the preprocessing steps for each email. You should
    #  complete the code in processEmail.m to produce a word indices vector
    #  for a given email.

    print('Preprocessing sample email (emailSample1.txt)\n')

    emailSample1 = open("./data/emailSample1.txt", 'r')

    contents = emailSample1.read()
    wordIndices = spamHelper.processEmail(contents)

    print('Word Indices:')
    for i in wordIndices:
        print('{:04d}'.format(i))

    input('\nPart 1 completed. Program paused. Press enter to continue: ')

    #  ==================== Part 2: Feature Extraction ====================
    #  Now, you will convert each email into a vector of features in R^n.
    #  You should complete the code in emailFeatures.m to produce a feature
    #  vector for a given email.

    print('\nExtracting features from sample email (emailSample1.txt)')

    # Extract Features
    featureVector = spamHelper.getEmailFeatures(wordIndices)

    print('Feature Vector stats:')
    print('Length: {}'.format(featureVector.size))
    print('Sum: {:}'.format(int(np.sum(featureVector))))

    input('\nPart 2 completed. Program paused. Press enter to continue: ')

    #  =========== Part 3: Train Linear SVM for Spam Classification ========
    #  In this section, you will train a linear classifier to determine if an
    #  email is Spam or Not-Spam.

    # Load the Spam Email dataset
    mat = io.loadmat('./data/spamTrain.mat')
    X = mat['X']
    y = mat['y'].astype(int).ravel()

    print('\nTraining Linear SVM (Spam Classification)')
    print('(this may take 1 to 2 minutes) ...')

    C = .1

    model = svm.SVC(
        C=C,
        kernel='linear')

    model.fit(X, y)

    print('Training Accuracy: {:.2f}'.format(
        100*model.score(X, y)))

    input('\nPart 3 completed. Program paused. Press enter to continue: ')

    #  =================== Part 4: Test Spam Classification ================
    #  After training the classifier, we can evaluate it on a test set. We have
    #  included a test set in spamTest.mat

    # Load the test dataset
    mat = io.loadmat('./data/spamTest.mat')
    Xtest = mat['Xtest']
    ytest = mat['ytest'].astype(int).ravel()

    print('\nTraining Accuracy: {:.2f}'.format(
        100*model.score(Xtest, ytest)))

    input('\nPart 4 completed. Program paused. Press enter to continue: ')

    # ================= Part 5: Top Predictors of Spam ====================
    #  Since the model we are training is a linear SVM, we can inspect the
    #  weights learned by the model to understand better how it is determining
    #  whether an email is spam or not. The following code finds the words with
    #  the highest weights in the classifier. Informally, the classifier
    #  'thinks' that these words are the most likely indicators of spam.
    #

    # Sort the weights and obtain the vocabulary list

    coefficients = model.coef_.ravel()

    sortedPosition = np.argsort(coefficients)
    sortedPosition = np.flip(sortedPosition, 0)

    vocabList = vocab.dictionary
    inverseDictionary = {v: k for k, v in vocabList.items()}
    print('\nPosition, Coefficient, Word')
    for i in sortedPosition[:15]:
        print('{:04d}, {:.3f}, {}'.format(
            i,
            coefficients[i],
            inverseDictionary[i]))

    input('\nPart 5 completed. Program paused. Press enter to continue: ')

    #  =================== Part 6: Try Your Own Emails =====================
    #  Now that you've trained the spam classifier, you can use it on your own
    #  emails! In the starter code, we have included spamSample1.txt,
    #  spamSample2.txt, emailSample1.txt and emailSample2.txt as examples.
    #  The following code reads in one of these emails and then uses your
    #  learned SVM classifier to determine whether the email is Spam or
    #  Not Spam

    # Set the file to be read in (change this to spamSample2.txt,
    # emailSample1.txt or emailSample2.txt to see different predictions on
    # different emails types). Try your own emails as well!

    # read file
    filename = "./data/emailSample1.txt"
    email = open(filename, 'r')
    contents = email.read()
    email.close()

    # Extract Features
    wordIndices = spamHelper.processEmail(contents)
    featureVector = spamHelper.getEmailFeatures(wordIndices)

    # Expand features
    featureVector = featureVector[np.newaxis, :]

    prediction = model.predict(featureVector)

    print('\nProcessed file: \"{}\"'.format(filename))
    print('Spam Classification: {}'.format(prediction[0]))
    print('(1 indicates spam, 0 indicates not spam)')

    input('\nPart 6 completed. Program completed. Press enter to exit: ')

if __name__ == '__main__':
    main()