Computationally Efficient Approximation Algorithm of Dominance Based Rough Set Approach

Abstract

Information systems contain preference ordered decision attributes now a day. To deal with them Dominance based Rough Set Approach (DRSA) is used. Two important concepts in DRSA are upper and lower approximations, both of them are computationally heavy algorithms. Reduction in its computational complexity helps in many advance applications of DRSA. Our proposed algorithm reduces its complexity by making changes in conventional procedure. Our solution stops the algorithm from precomputing the same thing from scratch. We tested our proposed solution on 5 datasets from UCI. There is 50% reduction in complexity of comparison step, which will lead to overall reduction of computational complexity for calculating lower and upper approximations.