A dendrite method for cluster analysis

✍ Scribed by Calinski, T.; Harabasz, J.

• Book ID: 111872877
• Publisher: Taylor and Francis Group
• Year: 1974
• Tongue: English
• Weight: 1004 KB
• Volume: 3
• Category: Article
• ISSN: 0361-0926
• DOI: 10.1080/03610927408827101

✦ Synopsis

A method f o r identifying c l u s t e r s of points i n a m u l t idimensional Euclidean space is described and its application t o taxonomy considered. It reconciles, i n a sense, two d i ff e r e n t approaches t o the investigation of the s p a t i a l relationships between the points, viz., the agglomerative and the divisive methods. A graph, the s h o r t e s t dendrite of Florek e t a l . ( l 9 j l a ) , is constructed on a nearest neighbour basis ana then divided i n t o c l u s t e r s by applying the c r i t e r i o n of minimum within-cluster sum of squares. This procedure ensures an effective reduction of the number of possible s p l i t s . The method may be applied t o a dichotomous division, b u t is perf e c t l y suitable a l s o f o r a global division i n t o any number of clusters. An informal indicator of the "best number" of clust e r s i s suggested. It i s a "variance r a t i o c r i t e r i o n " giving some insight i n t o the structure of the pointa. The method is