采用偏序集层次聚类的多准则决策方法

Multi-criteria-decision-making method based on hierarchical clustering of poset

为有效展示决策结果的层次关系,采用偏序集理论与方法,结合权重顺序即偏好序优化层次聚类分析,构建基于层次聚类的多准则决策方法.利用该方法,对某商业银行提供的贷款数据进行分析,通过哈斯图节点层次变化,能较为准确发现潜在风险客户,较早避免风险发生.研究结果表明:该方法克服了传统多准则决策研究中只关注最优解或满意解,忽略解集之间层次结构的弊端,突破传统方法过于依赖权重数值的瓶颈,能为决策者提供更好的决策依据.

In order to effectively display the hierarchical relationship of the decision result set,a multi-criteria decision-making method based on hierarchical clustering is constructed by adopting the theory and method of poset,combined with the weight order,that is,the preference order optimization hierarchical clustering analysis.By analyzing the loan data provided by commercial banks,through the change of the node level in the hasse diagram,potential risky customers can be found more accurately,and the risk can be avoided earlier.The results of study show that this method overcomes the drawback of traditional multi-criteria decision making research that only focuses on the optimal solution or satisfactory solution and ignores the disadvantages of the hierarchical structure between the understanding sets,and breaks through the bottleneck of the traditional method that relies too much on the weight value,and can provide better decision-making basis for decision-makers.