A Rough Set-Based Heuristic Algorithm for Attribute Reduction

It is well known that finding the shortest reduct is NP hard. In this paper, a novel heuristic algorithm based on relative attribute dependency in rough set is proposed for attribute reduction in decision information systems. To find an optimal reduct, we use cardinality attributes as the heuristic. The algorithm to find optimal reduct of condition attributes based on the relative attribute dependency is implemented by using C language. Compared with the positive region calculating algorithm, the new algorithm calculates the relative attribute dependency degree, instead of generating positive region. The time of complexity of new algorithm is O(|A|*|A|*|U|*log|U|) , where |A| is the number of condition attributes, and |U| is the number of objects in the decision information system. Experiments show that the new algorithm is more efficient on attribute reduction in decision information system.