Laboratory 5
Objective
• Implement Naïve Bayes classifiers from scratch
• Implement common evaluation metrics
(Note: “Implementation from scratch” means “not relying on any pre-implemented machine learning libraries”, but libraries such as NumPy, Pandas and SciPy can be used.)
Naïve Bayes classifier
Review the Naïve Bayes classifier theory and implementation technique in Lecture 5. Develop a Naïve Bayes classifier as an email filter in Python. Namely, the classifier predicts whether emails are ordinary or adverts.
Dataset: Files “wordsList” and “classList” (available in the datasets folder of this assignment on Moodle) The wordsList file contains 72 pre-processed emails. Each line is a list of words extracted from each email. The classList file contains the class labels that indicating whether the emails are ordinary or adverts (0 for ordinary and 1 for adverts).
Requirements
• Use stratified sampling to randomly select 66 out of 72 lines for training and the remaining 6 lines for test. Return the classification probabilities of these 6 records.
• The Naïve Bayesian classifier must account for multiple occurrences of words and implements techniques to overcome the numerical underflows and zero counts.
• Compute the TP/TN/FN/FP and plot the ROC for your classifier.
