Research On The Comprehensive Evaluation Based On The Rough Set And The Fuzzy Set Theory

Many objects in the reality need to be evaluated comprehensively. The general evaluation method is based on statistical theory and analyzes the current data. Generally, an improper evaluation method usually causes unfair results.Therefore, the study of comprehensive evaluation method is of great importance in theory and practice. This thesis, based on the fuzzy set theory, attempts to discuss the fuzzy comprehensive evaluation method and evaluate the objects in a quantitative way. This analysis, based on the rough set theory, also tries to use the data mining to evaluate the objects qualitatively. The paper applies this comprehensive method to evaluate the teaching quality and gets a rather satisfactory result. The main procedures of the body is conducted as the following:Firstly, the body is concerned with the overall evaluation process of the data mining. Contrasted with the traditional evaluation method, which usually deals with the current data, the paper introduces the data warehouse in the data mining and combines the current data with the historical data and analyzes them comprehensively. And then the overall process of the analysis is subdivided into several procedures: building data warehouse upon the evaluated data, data drawing, data pre-processing, comprehensive evaluation analyzing and output.Secondly, the paper studies the data pre-processing, including filling those null values and analyzing the discrete process of attribute value. The unproved ROUSTIDA algorithm is adopted to deal with the former as well as the fuzzy cluster with the latter.Thirdly, the body is concerned with the study of the data outcome after pre-processing. The fuzzy comprehensive evaluation method is used to describe each object quantitatively.Finally, the last part of the paper is to study the data-mining algorithm based on the rough set theory. Relying on the combination of the entropy and therough set theory, a relativity theory is put forward that the condition attribute is relative to the decision, which is applied to the searching algorithm of the attribute reduction and in the end each object is described qualitatively.