Naive Bayes Tutorial - Project 0 (Intro to Machine Learning)

What does the Naive in Naive Bayes theorem mean?

It assumes that all the features are independent of each other.

What is a prior (with regards to Bayes theorem)?

Probabilities that we are aware of.

What is a posterior (with regards to Bayes theorem)?

Probabilities we are looking to compute using the priors.

What is Bayes Theorem formula?

P(A|B) = ( P(B|A) * P(A) ) / P(B)

What is the Naive Bayes algorithm for multiple features?

P(y | x1, ..., x2) = (P(y) * P(x1, ...., xn | y)) / P(x1, ...., xn)

What does "recall/sensitivity" mean when evaluating a model?

Tells us what proportion of messages that were actually spam were classified by us as spam. The formula is: True positives / (True positives + False negatives)

What is the F1 score?

The weighted average of the precision and recall scores. The score can range from 0 to 1. 1 is the best possible score

What is Gaussian Naive Bayes good for?

Suitable for classification of continuous data as it assumes the input data has a Gaussian/normal distribution

What is Multinomial Naive Bayes good for?

Suitable for classification with discrete features such as word counts for text (e.g. spam or not) classification

What is the Bag of Words concept?

Bag of words takes a document of words and counts the frequency of every word.

What does "accuracy" mean when evaluating a model?

Number of correct predictions / total number of predictions

What does "precision" mean when evaluating a model?

Tells us what portion of messages that were classified as spam were actually spam. The formula is: True positives / (True positives + False positives)