Clustering validity checking methods

TLDR

Clustering results validation is an important topic in pattern recognition. The paper reviews clustering validity approaches, highlights under‑addressed issues, and proposes future research directions. The authors present internal and external criterion‑based validity methods and review relative criterion‑based approaches. Results from an experimental study using widely known validity indices are discussed.

Abstract

Clustering results validation is an important topic in the context of pattern recognition. We review approaches and systems in this context. In the first part of this paper we presented clustering validity checking approaches based on internal and external criteria. In the second, current part, we present a review of clustering validity approaches based on relative criteria. Also we discuss the results of an experimental study based on widely known validity indices. Finally the paper illustrates the issues that are under-addressed by the recent approaches and proposes the research directions in the field.