A hierarchy reduct algorithm for feature subset selection
Binbin Qu, Yansheng Lu · 2005
Practical machine learning algorithms are known to degrade in performance when faced with many features. Feature subset selection is the problem of choosing a small subset of feature that is necessary and sufficient has been proposed. However, the problem of generating a minimal reduct has been proved to be NP-hard. We propose an algorithm based on rough sets theory. The algorithm adopts approximation quality concept and hierarchy structure. The validity and feasibility of the algorithms are demonstrated by experiments. Experiment shows that the algorithm can select a better subset of features quickly and effectively.