Research On Data Reduction Algorithm Based On Rough Set

With the rapid development of today’s information, how to excavate implicit anduseful information from massive dataset has became the hot issue of the current researchin the area of date mining. The rough sets theory is an effective data mining tool forprocessing vague and imprecise knowledge, has been generally used in medical diagnosis,pattern recognition, and other fields. Attribute reduction is the focus of rough sets, it is aform of data statute. The basis idea of the attribute reduction algorithm is on the premiseof maintaining the ability of classification, to delete unnecessary attributes and reduce thedimensionality of dataset. This paper studies the classical rough set model and theextended rough set model for reduction algorithm on rough set, the purpose is to seek arapid and effective attribute algorithm. This paper analyzes the inadequacy of theattribute reduction based on mutual information and consistency criterion. The mainresearch contents are as follows:(1) In order to solve the inefficiency of core algorithm based on classical positiveregion, this paper put forward a new core algorithm based on simplified positive region,the experimental results show that the new algorithm is faster.(2) The efficiency of attribute reduction algorithm based on consistency criterion isstudied. This paper redefines the importance degree of attribute and the method forcalculating core from the perspective of the object consistency. On this basis, animproved attribute reduction algorithm based on consistency criterion is proposed.Selecting suitable consistency parameter ε and datasets for the experiments, proved thatthe improved algorithm is faster.(3) Research on rough sets in cardiovascular diagnosis. This paper analyzes thebasic characteristics of the cardiovascular data and introduces the pretreatment method.This paper processes cardiovascular diagnosis instance by using attribute reduction basedon consistency criterion and mutual information, and the results show the effectivenessof the attribute reduction based on rough sets.