A New Heuristic Feature Selection Algorithm Based on Rough Sets

A heuristic algorithm of reduct computation for feature selection in data mining is proposed in this paper, which aims at reducing the number of irrelevant and redundant features. This algorithm is based on the modified dependency degree formula. The advantage of this algorithm is that it can find the optimal reduct set for feature selection with less time complexity in most cases. To test the validity and generality of this algorithm, experimental results with 7 data sets from UCI Machine Learning Repository are given.