Entropy Based Bayes' Rule for Coping Dimensionality Reduction in Predictive Task of Data Mining

An integrated framework is developed in this work for reduction of dimensionality and prediction of the class label of unseen sample by using the best attribute of entropy measure and Bayes predictive rule. Dimensionality reduction is one of the fundamental problems of pre-processing phase of knowledge discovery in databases. Hence, for realizing the importance of dimensionality reduction across domains, a filter like approach through entropy measure is developed in this work. The framework is evaluated thoroughly through a few benchmark datasets obtained from University of California, Irvine(UCI) Machine Learning repository and then compared with canonical Naive-Bayesian classifier.