关键词: 特征选择 数据挖掘 KDD 粗糙集

Abstract:

In this paper, based on the Minimum Different List （MDL） of the rough set theory, we build a Simple Different Matrix （SMD） which is similar to MDLusing the incremental representation. From the approximate reduction set of SDM, we implement the forward selection of the attribute subspace. Finally,a hybrid feature selection algorithm based on the rough set theory has been proposed, and the result shows that the efficiency and accuracy of feature selection have been improved greatly by this approach. What is more, the algorithm does well in the preprocessing of data mining.

Key words: （feature selection, data mining, KDD, rough set）