基于新直觉模糊相似度的聚类方法

针对现有直觉模糊聚类方法大都未考虑属性(指标)权重,计算过于复杂且计算结果为实数的问题,提出一种基于新直觉模糊相似度的聚类方法,计算结果为直觉模糊数,运用直觉模糊熵得到属性权重,构造了一种考虑属性权重的直觉模糊相似度公式,得到直觉模糊相似矩阵,设计了风险参数,决策者根据自己风险偏好选择风险参数进行聚类.最后通过算例验证了所提出方法的可行性和合理性.

Clustering method based on new intuitionistic fuzzy similarity degree

Most existing clustering methods to intuitionistic fuzzy sets do not take the weight of attributes into account or
need too complex calculation and the result is real number. Therefore, a clustering method based on the intuitionistic fuzzy
similarity degree with the result being intuitionistic fuzzy number is proposed. Firstly, the weight of attributes is obtained
by utilizing entropy for intuitionistic fuzzy sets. A formula is proposed to derive the intuitionistic fuzzy similarity degree
between two intuitionistic fuzzy sets and an approach is developed to construct an intuitionistic fuzzy similarity matrix. A
risk parameter is designed and decision makers can cluster according to their own risk preference. Finally, an example shows
the feasibility and validity of this method.