Based On Rough Set And Fuzzy Clustering Decision Support System Design And Implementation Of Meteorological Disasters

The decision-making is a process to select the program which has the largest utility by the decision-maker through their domain knowledge. Considering that the amount of information is huge and the knowledge is showing greater uncertainty, at the same time, the decision-making is facing a complex environment, so, the research for intelligent decision-making methods and decision-making system is of great significance.In the field of Data Mining and Knowledge Discovery, fuzzy set and rough set theory have been proved to be very effective. The integration of the two theories can help us to build up a fuzzy-rough set based information system. With the advantage of the combination of the two theories, we can do the knowledge discovery in large amounts of data effectively with lower system complexity.This paper studied and implemented an intelligent decision-making system for meteorological disasters based on rough sets and fuzzy clustering, as follows:(1) This paper did the research of the fuzzy clustering algorithm based on point density: Classic fuzzy clustering algorithm can get a good classification of fuzzy things on the real-life, but for the classification with different information density, the algorithm is less than ideal. This paper raised an adjustment factor based on the point density to improve the original Euclidean distance metric, which made the algorithm much more suitable for the data with different information density. With the algorithm, we can get the degree of membership of each sample, which can build up a fuzzy information system.(2) This paper did the research of the decision tree induction algorithm based on fuzzy-rough set:We have to build the fuzzy decision tree after getting the fuzzy information system. The rough set theory can be used to reduce the dimension of the information system, so this paper applied the rough set theory to the fuzzy ID3algorithm, and did the decision tree induction using the degree of importance of the fuzzy attribute in the fuzzy-rough set theory, instead of the fuzzy classification entropy, which reduced the number of classification rules but with high accuracy.(3) Finally this paper applied the fuzzy clustering and the decision tree induction algorithm to the decision-making system for meteorological disasters, and did the verification for the system with actual data.