n South African Computer Journal - Naive Bayesian classifiers for multinomial features : a theoretical analysis : pattern recognition special edition
|Article Title||Naive Bayesian classifiers for multinomial features : a theoretical analysis : pattern recognition special edition|
|© Publisher:||South African Computer Society (SAICSIT)|
|Journal||South African Computer Journal|
|Author||H.O. Van Dyk and E. Barnard|
|Publication Date||Jun 2008|
|Pages||37 - 43|
|Keyword(s)||Bernoulli process, Bernoulli trial, Binomial, Correlation, Error estimate, Modied log likelihood, Multinomial, Naive Bayesian and Variance|
We investigate the use of naive Bayesian classifiers for multinomial feature spaces and derive error estimates for these classifiers. The error analysis is done by developing a mathematical model to estimate the probability density functions for all multinomial likelihood functions describing different classes. We also develop a simplified method to account for the correlation between multinomial variables. With accurate estimates for the distributions of all the likelihood functions, we are able to calculate classification error estimates for any such multinomial likelihood classifier. This error estimate can be used for feature selection, since it is easy to predict the effect that different features have on the error rate performance.
Article metrics loading...