The Fuzzy Clustering is a useful method for identifying the fundamental structures of a given homogenous dataset and is being applied in wide variety of scientific disciplines. Fuzzy clustering is a technique to cluster the high dimensional and noisy data where mining or grouping of the data is needed. This includes many real time applications such as clinical, surgery data, sales of a company and weather prediction. In this paper, the analysis of clustering algorithms like Fuzzy Particle Swarm Optimization (PSO), Fuzzy C-Means (FCM), and Hybrid FCM-PSO for two different types of dataset (without missing values and with missing values) related to heart disease diagnosis are implemented and evaluated. Initially the uncertain dataset is filled using Most Common Method. This algorithm is used for further optimize the output of FCM the Fuzzy PSO is applied for both the dataset. Finally the Hybrid FCM - PSO is applied and cluster centers are obtained which is in turn compared with clustering centers derived from FCM. The complete, comprehensive and effective clustering solution is measured with compactness and probability of clusters. The analysis proves that the Hybrid FCM - PSO is resulting with more accurate cluster centers.
Volume 11 | 02-Special Issue
Pages: 1600-1608