Research Of Algorithm For Attribute Reduction Based On Rough Set

With the advent of the information age, the amount of data is increasing sharply, and people hope that the data analysis tools can automatically obtain potential and useful knowledge from mass data without processing of human. In this paper, the problem of Extension of rough set under incomplete information and heuristic algorithm for attribute reduction on Rough set was studied, the main research work are in the following two aspects:(1) Extension of rough set under incomplete information was studied. The traditional rough set theory is mainly used to deal with a complete information system which only has discrete attribute values. But in reality, due to error of data measurement, loss of data during data transmission, human error, and many other reasons, the final information systems are often incomplete. Therefore, we study three rough set model based on tolerance relation, similarity relation, characteristic relation, and analyze numerical characteristic of incomplete decision system, and define the degree of incompleteness and the degree of completeness of the incomplete information system, and propose the idea of separating a complete sub-information system from the incomplete information system in order to parallel processing.(2) Heuristic algorithm for attribute reduction of decision system was studied. According to the concept of change rate, a new attribute importance measure method is defined from the viewpoint of information theory, and the corresponding algorithms for attribute reduction based on mutual-information change rate are proposed. At last, the experimental results show that the algorithms can effectively reduce the decision system.