Study Decision-making Approaches Based On Dominance Relation Rough Sets In Incomplete Information System

Classical rough set theory is based on equivalence relation, assuming thateach sample objects have definite attribute values. But in the actual data miningproblems, the data to be processed is always incomplete because of the influenceof various factors, which will affect the data analysis. Therefore, it has becomeone of the important research subjects that rough set theory is applied to acquireknowledge from incomplete information system at present.Dominance relation rough set theory has been extended by many relatedscholars, which is applied in the incomplete information system. This paper isaimed at further studying all kinds of extended dominance relation rough set inincomplete information system. Comparative analysis of the existing dominancerelation rough set model, Found that the model still exist some disadvantages: theexisting dominance relation rough set direct ruled out them without considerationthat attribute values of two objects with same attribute are all null or one of thetwo object’s attribute value is null and the other one is the middle value(Theseattributes is called unable compared attribute). The decision results will have agreat deviation with the actual result that the existing dominance relation roughset is applied to make decisions when the above situation occurs frequently.According to the above problem, this paper introduced the concept of thecritical value in order that incomplete information system is divided into twocases to discuss according to the number of unable compared attributes. The firstcase, building the imp-roved and limited dominance relation rough set modelwhich retains the advantages of limited dominance relation under the conditionthat the critical value is greater than or equal to the critical value. The secondcase, making a paired comparison table whose attribute value is interval number, Building a variable precision and dominance relation rough set model combinethe expansion theories of dominance relation rough set and variable precisionrough set, giving methods of attribute reduction and extraction of decision rules,setting the rule confidence threshold value of the variable precision rough set(VPRS) is variable parameter values, gaining a decision-making approach basedon variable precision and dominance relation rough set. The decision method isproposed in this paper Retained the advantages of the existing dominancerelation rough sets, solved the problem that a large number of unable comparedattributes is processed imperfect in incomplete information system, which willproduce important influence to decision result. Finally, this paper took theShenzhen municipal government purchasing supplier selection for example inorder to verify the practicability of this method.