A uniform approach to multidimensional scaling (Q1203098)

Given the distances between a collection of points in some metric space one can try to reconstruct the configuration of the points. When the distances themselves are subject to errors this reconstruction process is called multidimensional scaling. Two strategies that have been applied are interval scaling (metric) and ordinal (nonmetric). This paper shows how these can be regarded as opposite ends of a continuum of scaling methods. The algorithms described in the paper are compared to ones that are included in the SSPS-X package.