Cluster Analysis and Comparative Study of Different Clustering Performance and Validity Indices

One of the most well-known problems in data mining is clustering. Clustering, an unsupervised classification technique, entails the identification of object groups characterized by intra-group similarities and inter-group dissimilarities. Most clustering algorithms are highly sensitive to input parameters. The validation of the results obtained by clustering algorithms is an essential part of the clustering process. Therefore, evaluating the outcome of clustering algorithms is of great importance. In this study, real-life event data are clustered using different versions of CLARA and WARD algorithms, and clustering quality is measured and evaluated using clustering validity and performance indices. A presentation of two external and fifteen internal clustering indices is given in this context. Additionally, it also proposes developing appropriate measures to find an ideal clustering algorithm and provide insights into the performance and quality of the dataset.