Based On Rough Set Theory, Data Mining Method And Its Application

The data mining technique is a combination of machine learning, database and statistical theory. It has a widely foreground of application. Rough set theory, introduced by Z.Pawlak in the early 1980s, is a mathematical tool used for dealing with vagueness and uncertainty. In recent years it has received great attention of researchers around the world and has been successfully applied in many areas, such as artificial intelligence, knowledge discovery in database, pattern recognition, classification, expert system and fault diagnostication . This thesis studies rough set theory and its application.Data mining technique and rough set theory method are summarized in this thesis. Data mining is classed based on knowledge. The rough set theory's basic concepts and properties are summarized, attribute reduction and attribute value reduction's algorithms are also summarized. In the phase of data mining, there are three methods: standard attribute reduction algorithm that based on data analysis, heuristic attribute reduction algorithm that based on frequency of attribute and reduction based on discernibility matrix.Along with the rapid increase of data, the incremental data mining has raised wide concerns. The rough set incremental data mining is solved by dynamic reduction. The method of rough set incremental data mining, as we introduce, by theory analysis, it is an effectual way to deal with incremental data. At the same time, the application of dynamic reduction avoids calculating |U|×|U| matrix, and saves time of calculating reducts, thus extending the application scope of rough set data mining.The rule extraction is core of data mining. A heuristic attribute reduction finding algorithm is proposed. The algorithms improve rapidity of data mining, and also realize rule extraction automatically. The data reasoning of rough set depends on the form of rule. Traditional Data mining technique usually extracts rule from conditional attribute to decision system, such as Râ†'D. In this thesis, according to rule reasoning concepts in rough set theory, a decision attribute oriented information fusion method is presented in the consistent decision system.The main results are as follows: 1.As to inconsistent decision information system, a kind of new discernibility function is constructed. The influence degree of conditional attribute to decision system is described by probability distributing function, and maximal distributing set is obtained from the inconsistent decision information system.2.As to limit of general algorithm, reduction algorithm is improved, which based on discernibility matrix and significance of attributes.3.As to significance of frequency of attribute, a heuristic attribute reduction algorithm is improved, which based on frequency of attribute.4.As to incremental data mining, a method of rough set incremental data mining is presented, which based on dynamic reduction.5.As to rule reasoning of rough set, a decision attribute oriented information fusion algorithm is presented.6.Data mining system is designed and implemented, which based on rough set theory.