该matlab代码实现了“综合多分支分裂式层次化聚类”-Matlab implementation of Omnibus Multi-Branching Divisive Hierarchical Clustering (OMBDHC)

“A competitive omnibus performance criterion for divisive multi-branching

hierarchical clustering” by Soeria-Atmadja et al. (submitted). Note that OMB-DHC package

is implemented in Matlab code and therefore demands the Matlab core program, as well as the

Statistics toolbox of Matlab.

Divisive hierarchical clustering (DHC) has emerged as a promising alternative for the

identification of sub-structures in multivariate data. Of particular interest is multi-branching

DHC, which allows flexible numbers of subclusters at each hierarchical level. One poorly

explored issue in multi-branching DHC concerns the actual importance of the two

performance criteria (and their associated algorithms) to automatically create clusters and

select number of clusters, respectively. Another interesting but hitherto unexplored issue is

the possibility to employ a single omnibus performance criterion that guide