针对区间数模糊c均值聚类算法存在模糊度指数m无法准确描述数据簇划分情况的问题，对点数据集合的区间Ⅱ型模糊c均值聚类算法进行拓展，将其扩展到区间型不确定数据的聚类中。同时，分析了区间数的区间Ⅱ型模糊c均值聚类算法的收敛性，以确定模糊度指数m1和m2的取值原则。基于合成数据和实测数据的仿真实验结果表明：区间数的区间Ⅱ型模糊c均值聚类算法比区间数的模糊c均值聚类算法的聚类效果好。%In the fuzzy c-means clustering method for interval-valued data, the fuzzifier is responsible for clustering performance. However, it is impossible to accurately confirm the fuzzifier with a single value because of the uncertainty dispersion of the dataset. In this paper, we extend the IT2 FCM clustering method for point data to that for interval data, and exploit the differences between these two clustering methods by comparing their iterative processes. The iteration process of the KM algorithm is discussed and the selection rules for fuzzifiers for IT2 IFCM clustering method is provided in this paper. The validity of the proposed clustering method is investigated and compared to the IFCM clustering methods for synthetic and real interval-valued datasets. Computational results verify the validity of the proposed method.
展开▼
