基于区间数贴近度的最大树聚类算法研究

针对特征指标和权重指标均为区间数的多指标信息聚类问题,在传统的数值信息最大树聚类法的基础上,给出了基于区间数的多指标信息聚类问题的一种具体算法.在此过程中,加权相似矩阵的提出克服了属性权重对聚类产生的双重影响;并通过引入相对贴近度的概念,对相似区间进行排序.最后,通过实例验证了此算法的有效性.

A Study of the Maximum Tree Clustering Algorithm Based on Interval Number Approach

With respect to the problems of which both the characteristic values and weights of indices are uncertain interval numbers, based on the traditional maximum tree clustering method, this paper provides a clustering algorithm based on the clustering problem of multiple indicators. In this process,the proposed weighted similar matrix overcomes the double impact which attribute weights generated to the clustering results,and we sort the similar interval numbers by introducing the conception of the relative approach degree. Finally, we demonstrate the effectiveness of this algorithm by an example.