Cluster analysis beingun-supervised knowledgemethod where the objects are organized into groups based on their similarities. A hard partition induces the completetask of assigning data points to one cluster. Fuzzy clusteringpermitsaddition of continuing memberships of data to clusters. Fuzzy clustering simplifies partition based clustering approaches by permitting a discretedatapointto be categorized into two or more clusters. Fuzzy C-Means (FCM) approachdoes clusters by repetitivelyprobing for a set of fuzzy clusters. A fuzzy clustering approach was implemented to delineate nearly 13 drugs which are known to inhibit GSK-3 beta in a computational virtual screening study. Clusterability of the dataset was investigated by Hopkins statistic. Sum of possible clusters for the dataset was estimated by few statisticmethods. Fanny and FCM methods and their respective validation procedures were carried out. Comparison of clusters from fanny and fuzzy c-means suggest that the FCM method is the best choice for the given dataset.
Volume 11 | 06-Special Issue
Pages: 1958-1966