The Research Of Data Mining Based On Rough Set Theory In Incomplete Information System

As a mathematical tool of dealing with fuzzies and uncertainties, rough set theoryhas it’s unique advantages. It can handle incomplete、 imprecise and uncertaininformation problem, and now data mining and pattern recognition has be combinedwith the success. Knowledge discovery or data mining select human preferences oruseful knowledge from the mass、missing、uncertain、fuzzy、noise and random data set.In the research of based on rough sets’ data mining, processing is the key to the wholedata mining. In data processing, we must do condition attributes reduction anddiscretization the data, because attribute reduction is one of the core content of rough set,so use the knowledge of rough set to do the data mining is a good choice. Fuzzy theoryis a tool has been used to deal with the fuzzy phenomenon and fuzzy concept, Inessence fuzzy mathematics dispose things is also uncertain, is fuzzy existing, fuzzymathematics and rough set has strong convergence and complementarity, how to rely onthese features, is the key to combined with the rough set model and the fuzzy modelsmoothly.This paper mainly discusses about combining the rough set knowledge and theknowledge sorting、the fuzzy knowledge、the SVM knowledge.（1） At first we studied of sorting0,1decision information system and laterto do the attribute reduction research based on discernibility matrix and heuristicalgorithm of attribute importance algorithm complexity compare with the classicrough set attribute reduction. Then we studied of sorting Complete informationsystem, later to do the attribute reduction research based on discernibility matrix andheuristic algorithm of attribute importance algorithm complexity compare with theclassic rough set attribute reduction. At last we studies of sorting incompleteinformation system and to do the attribute reduction.（2） Then we discuss the model based on the combination of fuzzy sets andrough sets, setting up a model based on tolerance about the fuzzy membershipfunction, then discuss attribute reduction of this model, in this paper use fuzzy setfuzzy characteristic evaluation to do attribute reduction substitute for rough set attribute reduction based on heuristic algorithm of attribute importance, and use fuzzyrelationship matrix demonstrates the rough set attribute reduction based ondiscernibility matrix dimension is can be reduction.（3） The last we simple research rouah set SVM model.