Study on a modified Fuzzy C-Means Clustering Algorithm
Shaohong Yin, Min Li · 2010
The traditional Fuzzy C-Means (FCM) Clustering Algorithm is widely used in Data Mining technology at present. It always adopts Euclidean Distance to measure the dissimilarity between objects. Accordingly the clusters with convex shapes could be generally discovered. But it is difficult to discover the clusters with irregular shapes, and also is more sensitive to the existence of noise and isolated points. In this paper, A modified Fuzzy C-Means Clustering algorithm based on Mahalanobis Distance algorithm, into which integrates the matriculated mind, is proposed. According to the final test and comparison on the data sets of Balance Scale and Artificial by Standard FCM algorithm, MatFCM for vectors algorithm and MatFCM for matrices algorithm respectively, it shows that the performance of our modified FCM clustering algorithm has a much better clustering result.