A comparative analysis of discretization methods for Medical Datamining with Naive Bayesian classifier

Naive Bayes classifier has gained wide popularity as a probability-based classification method despite its assumption that attributes are conditionally mutually independent given the class label. This paper makes a study into discretization techniques to improve the classification accuracy of Naive Bayes with respect to medical datasets. Our experimental results suggest that on an average, with minimum description length (MDL) discretization the Naive Bayes classifier seems to be the best performer compared to popular variants of Naive Bayes as well as some popular non-Naive Bayes statistical classifiers.