A Comparison Study between Various Fuzzy Clustering Algorithms

Clustering is the classification of objects into different groups, or more precisely, the partitioning of a data set into subsets (clusters), so that the data in each subset shares some common features. This paper reviews and compares between the two most famous clustering techniques: Fuzzy C-mean (FCM) algorithm and Subtractive clustering algorithm. The comparison is based on validity measurement of their clustering results. Highly non-linear functions are modeled and a comparison is made between the two algorithms according to their capabilities of modeling. Also the performance of the two algorithms is tested against experimental data. The number of clusters is changed for the fuzzy c-mean algorithm. The validity results are calculated for several cases. As for subtractive clustering, the radii parameter is changed to obtain different number of clusters. Generally, increasing the number of generated cluster yields an improvement in the validity index value. The optimal modelling results are obtained when the validity indices are on their optimal values. Also, the models generated from subtractive clustering usually are more accurate than those generated using FCM algorithm. A training algorithm is needed to accurately generate models using FCM. However, subtractive clustering does not need training algorithm. FCM has inconsistency problem where different runs of the FCM yields different results. On the other hand, subtractive algorithm produces consistent results.