Assessment of Cluster Tendency Methods for Visualizing the Data Partitions

Abstract

Clustering is a technique for grouping data objects based on similarity features that is widely used. Similarity metrics such as Euclidean, cosine, and others are used to generate similarity features. Traditional clustering approaches like k-means and other graph-based strategies are commonly used to find clusters. However, in order to determine the number of clusters, these approaches necessitate user intervention. Traditional clustering algorithms divide data without considering the number of clusters or cluster tendency beforehand. By using either k-means or graph-based clustering methods with an intractable ‘k' value set by the consumer, there is a risk of bad clustering performance. As a result, for prior knowledge of the number of clusters in clustering, it is essential to concentrate on cluster tendency methods. The various visual access tendency (VAT) methods for determining the number of clusters are presented in this paper.