Dimensionality Reduction Algorithm Based On Fuzzy Rough Sets

In recent years, data gathered by enterprises increase rapidly because of wide use of information technology ,but large volumes of data are not fully used and exploited after being collected, and databases which are full of useful information become data tombs that no body can make use of ,which is an enormous waste of resource. The rise and rapid development of technology of knowledge engineering makes drawing useful information from large volume data possible. But the high dimension of data still remains as a big obstacle for deducing rules and generating cases, and users must wait long for a output and the number of user that can work concurrently is limited.Fuzzy rough set theory is an effective tool for reduction of data dimension, but there are few dimension reduction algorithms that are based on fuzzy rough set theory so far. In this article, the author systematically summarizes current research and tendency of fuzzy rough set theory. Because different methods are used to deduce fuzzy rough approximations, there are mainly three kinds of fuzzy rough sets .The author also examine the method to fuzzify attributes in fuzzy rough set and present general method to fuzzify single attribute and composite attributes. After analyzing many dimension reduction algorithm based on rough set and fuzzy rough set theory,the author puts forward three effective algorithm based on fuzzy rough set theory : Decreaser ,GA_Reductor and Tree_Reductor. The Decreaser algorithm has a good computing comlexity. The GA_Reductor performance well , and the Tree_Reductor guarantee that the minimal reduct will be find without examining all possible subset of the set of all conditionalattributes.Railway invoices contains lots of information about customers and railway transportation. After careful analysis and conferring with experts on railway transportation,the author combines above algorithms to reduce the dimension of data in invoice database of GuangZhou Railway Group.The effect is considerable. And the attributes that affect customers' choice of transportation are found so a solid foundation for development of GuangZhou Railway Group Invoice Analysis System is built.