home | library | feedback

How many models are there?

B-Course considers only finite number of Naive Bayes models, but that number can be very big. The number of models depends on the number of variables we have.

Exponentially many models

For discrete data the number of different models is easy to state: if we have n predictor variables, we have 2 to the power of n (i.e 2 * 2 * 2 * .. (doing this n times) .. * 2 = 2n) possible models and B-Course tries to find the best of these. Notice that for a a relatively small n, this is already too big a number to be searched exhaustively since evaluating each model takes time too.


  B-Course, version 2.0.0
CoSCo 2002