PERFORMANCE ANALYSIS OF CLUSTERING ALGORITHMS BY USING DIFFERENT DATASETS

This study compares the performance of three of the top clustering algorithms—K-Means, HierarchicalClustering, and DBSCAN—by implementing them on different datasets. The study compares internal and externalmeasures of validation to determine how each of these algorithms would perform on different data distributionsand patterns. Partitional clustering, in which one cluster is allocated to every point of data, and fuzzy clustering,in which there exists overlap in membership between clusters, is studied in this research. The performancemeasurement uses two top measures: Silhouette Score, which quantitatively measures cluster quality, and DaviesBouldin Index (DBI), which measures intra-cluster coherence against inter-cluster distance. Dynamicvisualization methods are used in the study in order to give an interactive view of algorithmic performance,making it simpler for data scientists to choose the optimal clustering algorithms that suit dataset characteristics