An improved attribute reduction algorithm and its application in agriculture

The existing algorithms of attribute reduction based on discernibility matrix are complex in computation of the core, thus it is difficult to conduct attribute reduction on complex decision table about agricultural diseases and pests. If two or more attributes have the same importance value, it cannot solve the problem of choosing an attribute to add into the core. By simplifying the discernibility matrix, redefining the attribute importance function and introducing a concept of attribute influence, we propose an algorithm to solve the above problems and can obtain the minimum attribute reduction set. At last, the improved reduction algorithm is used to deal with a decision table of agricultural diseases and pests. The experimental result confirms the correctness and effectiveness of the algorithm. © 2014 by Binary Information Press.