Adaptive Mean-Linkage with Penalty: A new algorithm for cluster analysis

A novel cluster analysis technique, so-called Adaptive Mean-Linkage with Penalty algorithm (AMLP) is proposed. The method is based on a penalty concept applied to the Euclidian distance, which determines the dissimilarity among objects when clustering data. The implementation of this technique is straightforward and provides enhanced classification in our case studies. The proposed clustering procedure was applied to a dataset of compounds from the essential oil of plants acquired for classification purpose. The potentiality of this novel technique to distinguish each plant or group of plants according to the concentration levels of compounds in the essential oil has been validated. A free web tool is available.