PRTools Contents

PRTools User Guide



Naive Bayes classifier

    W = NAIVEBC(A,N)
    W = A*NAIVEBC([],N)
    W = A*NAIVEBC(N)


 A Training dataset
 N Scalar number of bins (default: 10)
 DENSMAP Untrained mapping for density estimation

 W Naive Bayes classifier mapping


The Naive Bayes Classifier estimates for every class and every feature  separately. Total class densities are constructed by assuming  independency and consequently multiplying the separate feature densities.

The default version divides each axis into N bins, counts the number of  training examples for each of the classes in each of the bins, and  classifies the object to the class that gives maximum posterior  probability. Missing values will be put into a separate bin.

This routine assumes continuous data. It may be applied to discrete data  in case all features have the same number of discrete values. For proper  results the parameter N should be set to this number.

If N is NaN it is optimised by REGOPTC.

Alternatively an untrained mapping DENSMAP may be supplied that will be  used to estimate the densities per class and per features separately.  Examples are PARZENM and GAUSSM. See also UDC.

See also

datasets, mappings, parzenm, gaussm, udc, qdc, parzenc, parzendc, regoptc, udc,

PRTools Contents

PRTools User Guide

This file has been automatically generated. If badly readable, use the help-command in Matlab.