基于决策者偏好投影寻踪模型的多属性决策法

Multiple Attribute Decision-making Method Based on Preference Information and Projecting Pursuit Classification Model

针对现有主观赋权法和客观赋权法的不足,提出了一种新的综合赋权方法——基于决策者偏好及投影寻踪聚类模型的综合赋权法。该方法运用投影寻踪聚类模型,把多属性决策问题中的高维数据转化到低维子空间,同时用自适应粒子群优化算法来优化投影指标函数和模型参数,获得了决策属性体系最佳投影方向和投影值,揭示了高维数据的结构特征。同时,也考虑了决策者对不同属性的偏好,使对属性的赋权达到主观与客观的统一。最后通过一个仿真实例说明了该方法的可行性与有效性。

In view of the shortage of the present subjective and objective assigning weight methods, a new combination assigning weight approach based on the decision-maker＇s preference and projecting pursuit classification model was proposed. Through applying projecting pursuit classification model based on adaptive particle swarm optimization algorithm in multiple attribute decision-making problems, the multi-dimension data of decision-making problem were easily changed into low dimension space and the multi-dimension data＇s structure feature could be discovered, Accordingly the optimum projection direction and the value of project function could be obtained. At the same time, this approach considered the decision-maker＇s preference information, too. The simulation results show that the proposed approach is effective and feasible.