Comparison of clustering performances of fuzzy C-means, possibilistic C-means and some fuzzy and possibilistic hybrid algorithms

İn this study, we comapred the performance of Fuzzy C-Means (FCM) and Possibilistic C-Means (PCM) and Possibilistic C-Means with Repulsion (PCMR), and their hybrids such as Fuzzy Possibilistic C-Means (FPCM), Possibilistic Fuzzy C-Means (PFCM), Possibilistic C-Means with Repulsion (PCMR) and Unsupervised Possibilistic Fuzzy C-Means (UPFCM). Four real data sets and eight synthetic data sets consists of various square, ellipse, circle and concave shaped clusters with some noises were used for testing the performance of the algorithms. Partition Entropy (PE), Partition Coefficent (PC), Modified Partition Coefficent (MPC), Xie-Beni (XB), Kwon, Tang-Sun (TS) and Fuzzy Silhouette (FS) were used as clustering validity indices for finding the optimal numbers of clusters in the analyzed data sets. According to the obtained results, UPFCM is proposed for partitioning of fuzzy and noisy large data sets because of its computational efficieny and success to find the optimal clustering results.