Naive Bayesian classifier for microarray data

Comparing with more sophisticated classifiers, the naive Bayesian classifier greatly simplifies learning by assuming that the attribute values are conditionally independent given the class. Although independence is a strong assumption, in practice naive Bayesian classifier often competes with other complex classifiers and naive Bayesian algorithm works well for classifying text documents. In this paper, we present our invented technique, called "attribute grouping" for data preprocessing. The naive Bayesian algorithm is implemented for classifying multiple gene expression patterns from microarray experiments. Results show that attribute grouping is very effective and that the naive Bayesian classifier becomes a suitable classification method for microarray data when the attribute grouping is used.