Naive Bayes

01 Jan 2016

Baye’s Classifier

• SImple rule to improve our predicition based on newly gathered evidence.

• Simple Classifier Here we have one feature discrete feature with data classifed into 2 classes depending on value of the feature. Now to predict for a future case given value of X we can convert to probabilties and can then calculate P(class|X) to be proptional to P(X|class) * P(class)

• Baye’s Rule We can now genralize the simplified classifier to many values of feature X and any number of classes.

Classifier for continious X

• For continious X we can first calculate the mean and standard deviation of the feature from our given data. Then we can assume one of the models for X. For example a gaussian model can be assumed.
• For each class we calculate the mean and standard deviation to arrive at a probabilty distribution for the same. Now using these we can calculate P(X|given class) and then calculate joint probabilties.

##Implementation

• Always split data into testing and training data. Testing on training data would give results way better than actual ones.
• Implementation in sklearn of GaussianNB