Consensus-based hierarchical agglomerative clustering in the context of weak orders

摘要

In this paper, we introduce a new hierarchical agglomerative clustering process in the setting of weak orders. This process is based on consensus measures induced by weighted Kemeny distances that associate a number between 0 and 1 to each subset of weak orders. Then, clusters are sequentially formed according to the consensus among the corresponding weak orders. The process is illustrated with the data obtained in an experiment carried out during the 2006 Public Choice Society election.