动态的模糊K-Modes初始化算法

模糊-Modes聚类算法针对分类属性的数据进行聚类，使用爬山法来寻找最优解，因此该算法对初始值较为敏感。为了克服该缺点，提出一种动态的模糊K—Modes初始化算法，该方法能够自动确定聚类数目，以及对应的聚类中心；而且能够应用于数值属性和分类属性相混合的数据集。该初始化算法可以有效地克服模糊K—Modes算法对初值的敏感性。实验的结果表明了该初始化算法的可行性和有效性。

Dynamic initial algorithm of fuzzy k-modes algorithm

Fuzzy k-modes clustering algorithm is efficient in clustering data sets with categorical domains. As a local searching optimization, however, it is sensitive to initial value. A method of dynamic initial algorithm of fuzzy k-modes algorithm was proposed, which could automatically give cluster number and modes of each cluster. It is also efficient in clustering data sets with categorical domains and domains with mixed numeric and categorical values, This initial algorithm could avoid the sensitivity of initial value of fuzzy k-modes algorithm. Experiment on data sets shows that the approach is effective.