Intelligent data cluster algorithm based on hierarchical FCM and Mahalanobis distance

Cluster method is used to categorize the same type of data points together. Fuzzy C-Means Clustering (FCM) has better cluster performance than traditional hard c-means clustering method. In the algorithm of FCM, the initial membership matrix of data is assumed in random normally, and the initial value affects the performance a lot meanwhile. In our previous study, a proposed hierarchical based FCM algorithm can give the proper initial value. In general, FCM based on Euclidean distance evaluate the membership value matrix to separate the data points into several clusters, but there are some data points can not be separated properly. Hence, this paper employs the Mahalanobis distance to improve the drawback of Euclidean distance function based method. The experiment results show that the proposed method can separate the data points well.