×

The fit of graphical displays to patterns of expectations. (English) Zbl 1080.62501

Summary: Graphical displays of multivariate data often clearly exhibit features of the expectations even though the data themselves are poorly fitted by the displays. Thus, it often occurs that ordinations and biplots that poorly fit the sample data still reveal salient characteristics such as clusters of similar individuals and patterns of correlation. This paper provides an explanation of this seemingly paradoxical phenomenon and shows that when many variables are analyzed, the common measure of goodness of fit of a lower rank approximation often seriously underestimates the closeness of the fit to underlying patterns. The paper also provides some guidelines on better estimates of the latter goodness of fit.

MSC:

62A09 Graphical methods in statistics
62H99 Multivariate analysis
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Besse, Ph.; Caussinus, H.; Fine, J.: Principal components analysis and optimization of graphical displays. Statistics 19, 301-312 (1988) · Zbl 0643.62046
[2] Besse, Ph.; De Falguerolles, A.: Application of resampling methods to the choice of dimension in principal component analysis. Computer intensive methods in statistics., 166-176 (1993)
[3] Caussinus, H.: Models and uses of principal component analysis. Multidimensional data analysis., 149-170 (1986)
[4] Caussinus, H.: Modèles fonctionels: quelques dévelopments et applications. Modèles pour l’analyse des données multidimensionnelles., 61-81 (1992)
[5] Caussinus, H.: Projections revelatrices. Modèles pour l’analyse des données multidimensionnelles., 241-265 (1992)
[6] Caussinus, H.: Modèles probabilistes et analyse des données multidimensionnelles. Journal de la société de statistique de Paris 134, 15-32 (1993)
[7] Caussinus, H.; Ferré, L.: Comparing the parameters of a model for several units by means of principal components analysis. Computational statistics and data analysis 13, 269-280 (1992) · Zbl 0800.62311
[8] Caussinus, H., Ruiz-Gazen, A., 1993. Projection pursuit and generalized principal component analysis, in New Directions in Statistical Data Analysis and Robustness. Basel, Birkhausen Verlag, pp. 35–46. · Zbl 0819.62052
[9] Caussinus, H., Ruiz-Gazen, A., 1995. Metrices for finding typical structures by means of principal component analysis, in Data Analysis and its Applications. Harcourt Brace, pp. 177–92.
[10] Eckart, C.; Young, G.: The approximation of one matrix by another of lower rank. Psychometrika 1, 211-218 (1936) · JFM 62.1075.02
[11] Ferré, L.: Improvement of some multidimensional estimates by reduction of dimensionality. J. multivariate anal. 54, 147-162 (1995) · Zbl 0845.62042
[12] Fine, J.: Modèles fonctionels et structurels. Modèles pour l’analyse des données multidimensionnelles., 21-60 (1992)
[13] Gower, J. C.: Statistical methods of comparing different multivariate analyses of the same data. Mathematics in the archeological and historical sciences., 138-149 (1971)
[14] Heo, M., 1996. On the fit of sample graphical displays to patterns in populations. Unpublished Ph.D. Dissertation University of Rochester, Rochester, NY.
[15] Householder, A. S.; Young, G.: Matrix approximation and latent roots. Amer. math. Monthly 45, 165-171 (1938) · Zbl 0019.14701
[16] Krzanowski, W. J.: Principles of multivariate analysis.. (1988) · Zbl 0678.62001
[17] Lingoes, J. C.; Schönemann, P. H.: Alternative measures for fit for the schönemann-carroll matrix Fitting algorithm. Psychometrika 39, 423-427 (1974) · Zbl 0295.92022
[18] Strauss, J. S.; Gabriel, K. R.; Kokes, R. F.; Ritzler, B. A.; Vanord, A.; Tarana, E.: Do psychiatric patients fit their diagnosis? patterns of symptomatology as described with the biplot. J. of nervous mental dis. 167, 105-113 (1979)
[19] Sibson, R.: Studies in the robustness of multidimensional scaling: procrustes statistics. J. roy. Statist. soc. B 41, 217-229 (1978) · Zbl 0389.62086
[20] Young, G.: Maximum likelihood estimation and factor analysis. Psychometrika 6, 49-53 (1940) · JFM 67.0489.01
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.