Interval estimation naïve Bayes

Abstract

Recent work in supervised learning has shown that a surprisingly simple Bayesian classifier called naïve Bayes is competitive with state of the art classifiers. This simple approach stands from assumptions of conditional independence among features given the class. In this paper a new naïve Bayes classifier called Interval Estimation naïve Bayes is proposed. Interval Estimation naïve Bayes is performed in two phases. First, an interval estimation of each probability necessary to specify the naïve Bayes is calculated. On the second phase the best combination of values inside these intervals is calculated using a heuristic search that is guided by the accuracy of the classifiers. The founded values in the search are the new parameters for the naïve Bayes classifier. Our new approach has shown to be quite competitive related to simple naïve Bayes. Experimental tests have been done with 21 data sets from the UCI repository. © Springer-Verlag Berlin Heidelberg 2003.