Monte Carlo studies in nonmetric scaling |
| |
Authors: | Ian Spence Forrest W. Young |
| |
Affiliation: | (1) Department of Psychology, University of Western Ontario, N6A 5C2 London, Ontario, Canada;(2) University of North Carolina, USA |
| |
Abstract: | In response to Arabie several random ranking studies are compared and discussed. Differences are typically very small, however it is noted that those studies which used arbitrary configurations tend to produce slightly higher stress values. The choice of starting configuration is discussed and we suggest that the use of a principal components decomposition of the doubly centered matrix of dissimilarities, or some transformation thereof, will yield an initial configuration which is superior to a randomly chosen one.This research was supported by the National Research Council of Canada (Grant No. A8351) and by the National Institute of Mental Health (Grant Nos. MH10006 and MH26504). The authorship order has been determined by Monte Carlo methods. |
| |
Keywords: | multidimensional scaling random rankings initial configuration |
本文献已被 SpringerLink 等数据库收录! |
|