Heuristic algorithm of attribute reduction in variable precision rough sets

Heuristic algorithms of attribute reduction in variable precision rough sets (VPRS) are proposed based on the analysis of the VPRS theory. The attribute significance is studied in terms of increment of attribute dependence, increment of mutual information, measure of degree of accuracy with degree of coverage and measure of uncertainty, which are taken as heuristic information in the heuristic algorithms. Moreover, the least reduction is given. The corresponding algorithms are implemented on the MATLAB. At last, a practical example is given to show the validity and practicability of the algorithm.