New Heuristic Attribute Reduction Algorithm Based on Rough Set

One of the main obstacles facing current data mining techniques is attribute reduction. This paper summarized advantages and disadvantages of current attribute reduction algorithm, and proposed a new attribute reduction method that is taken correlation degree as heuristic information; this method can remove not only irrelevant attributes, but also redundant attributes from the candidate attribute set. Theoretical analysis and experiment results demonstrate that on the premise of unchanged classification precision, the algorithm can obtain the best attribute reduce set effectively and efficiently.