Abstract: This paper presents a twophase algorithm for extraction of decision rules based on rough set and evidence theory. In the algorithm the thinking of reducing feature of rough set theory was used to get the important feature sets of each rule. Then on the basis of the thinking of evidence combination of evidence theorythe redundant features of the important feature sets was cut so as to get decision rules. The twophase algorithm presented in this paper simplifies the reducing of feature sets. And it is feasible for high dimensional data. The result of experiment shows that it can get fine decision rules.

Key words: rough set, evidence theory, extraction of decision rule