DataMining forMulti-Domain Decision-Making BasedonRoughSet Theory

Inthemodernworld, itisdifficult todiscover knowledge fromthesameinformation withmulti-purpose. Usually, itisonlypossible toconsider onepurpose atatime. If there havesameinformation corresponding withmulti-attribute decisions. Theusual approach isseparate thedecision attributes byinstinct orbyexpert knowledge inordertosimplify the purpose decisions. This, however, becomes moredifficult when there aremanymulti-attribute decisions. Ourproposed method usesthereduct process ofRoughSetTheorytoseparate multi-attribute decisions intomulti-domains tosimplify the analysis. Therealsoexistrelations amongthedomains. Generating decision tables by relation degreeavoidthe shortcomings ofhumandecision-making andarebetter for decision qualification. An empirical studyofmulti-domain decision-making intheinsurance market isusedtoillustrate the reduct process inourproposed method. Theresults demonstrate thatthereduct process canseparate multi-attribute decisions by attribute domains successfully. We takesevendecision tables thatwereoriginally drawnupbasedoninstinct orexpert knowledge andsimplify themintoonlythree decision tables. Thisenhances theprecision ofthedecision explanation. Decision tables generated according todomains aremorescientific than those derived bytraditional methods andmoreuseful fordata analysis.