论文信息 - Application of Resampling Methods to the Choice of Dimension in Principal Component Analysis

Application of Resampling Methods to the Choice of Dimension in Principal Component Analysis

This paper investigates the problem of the choice of dimension in Principal Component Analysis (PCA). PCA is introduced as a model; a loss function assessing the stability of the fit is considered. The choice of dimension then amounts to the minimisation of an expected loss which has to be estimated. This is achieved by resampling methods. Different bootstrap and jackknife estimates are presented. The behaviour of these estimates are investigated on artificial data and on real data. The resulting choices are confronted with those given by naive rules.

Ph. Besse | A. de Falguerolles

[1] H. Caussinus,et al. Principal components analysis and optimization of graphical displays , 1988 .

[2] A. Pousse,et al. Asympotic study of the multivariate functional model. application to the metric choice in principal component analysis , 1992 .

[3] Allan R. Wilks,et al. The new S language: a programming environment for data analysis and graphics , 1988 .

[4] J. Webb. Perturbation theory for a linear operator , 1967, Mathematical Proceedings of the Cambridge Philosophical Society.

[5] J. Daudin,et al. Stability of principal component analysis studied by the bootstrap method , 1988 .

[6] Philippe C. Besse. PCA stability and choice of dimensionality , 1992 .

[7] Pca stability studied by the bootstrap and the infinitesimal jackknife method , 1989 .

[8] B. Efron. Jackknife‐After‐Bootstrap Standard Errors and Influence Functions , 1992 .

[9] R. Beran,et al. Bootstrap Tests and Confidence Regions for Functions of a Covariance Matrix , 1985 .

[10] Tosio Kato. Perturbation theory for linear operators , 1966 .

[11] J. Leeuw,et al. Multidimensional Data Analysis , 1989 .

[12] G. C. McDonald,et al. Instabilities of Regression Estimates Relating Air Pollution to Mortality , 1973 .