Recommendations for validating hierarchical clustering in consumer sensory projects

Abstract

Choosing the proper hierarchical clustering algorithm and number of clusters is always a key question in consumer sensory projects. In many cases, researchers do not publish any reason why it was chosen a given distance measure and linkage rule along with cluster numbers. The reason behind this could be that different cluster validation and comparison techniques give contradictory results in most cases. A complex evaluation to define the proper clustering might be time-consuming and tedious. The paper introduces the clustering of three sensory data sets using different distance metrics and linkage rules for different numbers of clusters. The results of the validation methods deviate, suggesting that clustering depends heavily on the data set in question. Although Euclidean distance, Ward's method seems a safe choice, testing, and validation of different clustering combinations is strongly suggested.