Study On Rough Set-based Data Mining Techniques And Its Applications

Human capability to get useful information from database can't satisfy practical requirement in face of increasing data and database in information society. DBMS can't discover relationship rule in database and forecast prospective development tendency though it can achieve input, searching and safeguard efficiently. So we need a kind of technology and tool which can switch data to useful information by itself and intelligently.Demand is the mother of requirement. The development and combination of DBMS and machine learning in artificial intelligence contribute to the birth of new techniques-KDD in DB. It is the forum of No. 11 international artificial intelligence conference convoked in the United States that expound KDD first time in August, 1989. KDD is a crossing science; it involves many realms such as machine learning, mode identification, statistics, intelligent database, knowledge obtaining, data visible, high capability computing, expert system etc. Its connotation is extremely extensive, and the theory and technique are difficult as well. So the KDD technique aim for macro-database can't satisfy practical requirement. Then, ACM (Andeans Common Market) expound concept of Data Mining in 1995. It regards macro-database as the deposit bed of valuable information. We can mine useful information through effective knowledge discovery technology.The so-called data mining also called knowledge discovery, it is the process of providing potential and valuable knowledge or regular from macro-database. Date mining technique has all kinds of mode, such as association analysis, classification and forecast. Each mode has its own emphasis, among them, there are some already studied modes have much more research outcome, such as some methods in association rule mining, classification and forecast mode. But there are a few savants studying mining technique based on rough sets at present, and the corresponding research outcome is deficient as well. So data mining technique based on rough sets is worth studying. Rough sets is put forward for uncertain problem, it needn't quantitative depiction of some characters or attributes givenpreviously. Rough sets ascertains approximation domain of given problem from depiction set directly through indiscernibility relation. Thus we obtained interior rule of this problem.Data mining technique based on rough sets can be used for mining useful and interested knowledge. Therefore it solved the problem of more data and less information in enterprise. In a sense, rough sets is a kind of self-study mechanism, so we can solve the problem of knowledge obtained automatically by using rough sets. Accordingly, studying on data mining technique based on rough sets has important theoretical and realistic significance.Based on drawing lessons from expert experience, this paper systemically studies data mining technique based on rough sets. Previously people study compatible decision table principally, this paper not only studies compatible decision table, but also studies incompatible decision table and reasoning method of uncertain rule produced in mining. Thus data mining technique based on rough sets becomes more perfect.In fourth chapter, we studied two processes of data preparation, that is supplement and dispersing of decision table. In addition, this paper put forward a kind of method of data mining and reasoning by combining rough sets with cloud model.In sixth and seventh chapter thesis probes into applying of data mining based on rough sets in management field including selection of SCM partner and evaluation of enterprise credit. Partnership of SCM and evaluation of enterprise credit become very important under condition of market economy. Using rough sets-based data mining in these two fields not only can make full use of historical data and evaluation result, but also can improve efficiency and effect of selection and evaluation work.In sixth chapter, thesis constructs synthetic index system on selection and evaluation of partner, on the basis, it obtains classific...