Porównanie dokładności metody odległości probabilistycznej i podejścia zagregowanego w taksonomii

摘要

High accuracy of results is a very important aspect in any clustering problem t determines the effectiveness of decisions based on them. Therefore, literature proposes methods and solutions that aim to give more accurate and stable results than traditional clustering algorithms (e.g. k-means or hierarchical methods). Cluster ensembles (Leisch 1999; Dudoit, Fridlyand 2003; Hornik 2006; Fred, Jain 2002) or the distance clustering method (Ben-Israel, Iyigun 2008) are the examples of such solutions. Here, we carry out an experimental study to compare the accuracy of these two approaches.