Cluster validation is the process of evaluating the quality, reliability, and statistical significance of the groupings produced by a clustering algorithm. In unsupervised machine learning and data mining, where predefined ground-truth labels are typically unavailable, cluster validation assesses whether identified clusters reflect genuine underlying patterns rather than random noise or algorithmic bias. Evaluation techniques are generally divided into three categories: internal validation, which measures properties such as cohesion and separation using only the clustered data; external validation, which benchmarks the results against known reference labels; and relative validation, which evaluates different clustering configurations or algorithms to determine the optimal number of clusters and parameter settings.
