Lipshitz distance and hierarchical clustering
This note discusses the Lipshits distance between two different metrics (or definite) dissimilarities on a set. Given a metric space a universal lower bound is established for the Lipshits distance between the original metric on M and the metric defined by any hierarchical classification tree on M. Finally it is shown that single link clustering attains this lower bound.
|Department of Analysis, Algebra and Geometry [AM]|
Hazewinkel, M. (1995). Lipshitz distance and hierarchical clustering. Department of Analysis, Algebra and Geometry [AM]. CWI.