Exploration of common clustering methods and the behavior of certain performance indices

Statistical clustering is an exploratory method for finding groups of unlabeled observations in potentially high dimensional space, where each group contains observations that are similar to each other in some meaningful way. There are several methods of clustering, with the most common including hierarchical clustering, k-means clustering and model-based clustering. Agreement indices are quantitative metrics that compare two partitions or groupings of data. In this paper, we introduce three clustering methods and compare their results using different agreement indices, after being applied to Fisher’s iris data, a classic clustering benchmark data set