A Rough Set-Based Method for Constructing Simple Bayesian Classifier from Databases

In C 3I, for efficiently resolving some problems, such as threat-degree estimation, etc., the features that affect these objects should be correctly determined according to battlefield conditions. So a rough set-based method for constructing simple Bayesian classifier from databases was presented here. On the basis of the feature reduction algorithm based on rough set, this method takes synthetically into account the influence of the dependency of condition features and decision-making feature towards reduction, and the influence of the dependency among condition features towards reduction. By the dependency-based feature reduction, this method improves the independency limit among feature variables, so that the robust potential of simple Bayesian classifier is utilized and the performance of simple Bayesian classifier from databases is optimized. When those problems, such as threat-degree estimation, etc., were dealt with by this method in C 3I, obtained well experimental results.