论文部分内容阅读
基于Pawlak粗糙集的属性约简一般保持决策表的正区域不变,然而由于现实中不同用户对不同约简精度的需求,获取属性值的实际代价与个人偏好可能不同.针对决策者主观个人偏好、客观约简精度、获取属性值的实际代价和决策表各区域的误判代价等综合情况,提出新的约简算法,并讨论约简代价与约简精度间的关系.通过遗传算法,采用启发式方法搜索出局部最优约简子集.仿真实验表明,所提出的算法操作性强,更适合处理实际决策问题.
The attribute reduction based on Pawlak rough set generally keeps the positive area of the decision table unchanged, but due to different users’ different requirements for different reduction precision in reality, the actual cost and personal preference of obtaining attribute value may be different.For the decision maker’s subjective personal preference , The objective reduction precision, the actual cost of obtaining attribute values and the misjudgment cost of each decision table, a new reduction algorithm is proposed and the relationship between reduction cost and reduction precision is discussed.According to genetic algorithm, The heuristic method searches for the local optimal subset, and the simulation results show that the proposed algorithm is more operable and more suitable for solving practical decision-making problems.