作者: Shi Xu , Markad V. Kamath , David W. Capson
DOI: 10.1016/0167-8655(93)90127-Y
关键词:
摘要: Abstract A pragmatic stopping rule is proposed for helping a user select good partitions using agglomerative hierarchical clustering procedures. This combines the distance dissimilarity in dendrogram with squared-error criterion evaluation of all levels. The usefulness and limitation are demonstrated several data sets.