Data Mining for Multi-Domain Decision-Making Based on Rough Set Theory

In the modern world, it is difficult to discover knowledge from the same information with multi-purpose. Usually, it is only possible to consider one purpose at a time. If there have same information corresponding with multi-attribute decisions. The usual approach is separate the decision attributes by instinct or by expert knowledge in order to simplify the purpose decisions. This, however, becomes more difficult when there are many multi-attribute decisions. Our proposed method uses the reduct process of rough set theory to separate multi-attribute decisions into multi-domains to simplify the analysis. There also exist relations among the domains. Generating decision tables by relation degree avoid the shortcomings of human decision-making and are better for decision qualification. An empirical study of multi-domain decision-making in the insurance market is used to illustrate the reduct process in our proposed method. The results demonstrate that the reduct process can separate multi-attribute decisions by attribute domains successfully. We take seven decision tables that were originally drawn up based on instinct or expert knowledge and simplify them into only three decision tables. This enhances the precision of the decision explanation. Decision tables generated according to domains are more scientific than those derived by traditional methods and more useful for data analysis.