Distance Preservation and Index Adjustment in Hierarchical Clustering–HIERAG IV,a Turbo Pascal Program |
| |
| Authors: | R Henrion G Henrion |
| |
| Abstract: | Hierarchical clustering is a powerful tool for evaluating multivariate data as for instance about water samples. Results are illustrated by dendrograms which can be understood not only as means of classification but also as a kind of object display. It seems reasonable to preserve as well as possible original inter-object distances within the dendrogram. For the classical hierarchical algorithms a correction by index adjustment is proposed. This procedure may lead to more reliable results. Practical and simulated data examples are discussed. A formula for distance-updating in the case of relative distance preservation is constructed. Finally an advanced version of the program HIERAG in Turbo Pascal is listed. This program uses INTEGER-type variables for inter-object distances instead of REAL-type ones. |
| |
| Keywords: | |
|
|
