×

Regression of a data matrix on descriptors of both its rows and of its columns via latent variables: L-PLSR. (English) Zbl 1429.62027

Summary: A new approach is described, for extracting and visualising structures in a data matrix \(\mathbf Y\) in light of additional information BOTH about the ROWS in \(\mathbf Y\), given in matrix \(\mathbf X\), AND about the COLUMNS in \(\mathbf Y\), given in matrix \(\mathbf Z\). The three matrices \(\mathbf Z\)–\(\mathbf Y\)–\(\mathbf X\) may be envisioned as an “L-shape”; \(\mathbf X\) \((I\times K)\) and \(\mathbf Z\) \((J\times L)\) share no matrix size dimension, but are connected via \(\mathbf Y\) \((I\times J)\). A few linear combinations (components) are extracted from \(\mathbf X\) and from \(\mathbf Z\), and their interactions are used for bi-linear modelling of Y, as well as for bi-linear modelling of \(\mathbf X\) and \(\mathbf Z\) themselves. The components are defined by singular value decomposition (SVD) of \(\mathbf{X'YZ}\). Two versions of the L-PLSR are described-using one single SVD for all components, or component-wise SVDs after deflation.
The method is applied to the analysis of consumer liking data \(\mathbf Y\) of six products assessed by 125 persons, in light of 10 other product descriptors \(\mathbf X\) and 15 other person descriptors \(\mathbf Z\). Its performance is also checked on artificial data.

MSC:

62-08 Computational methods for problems pertaining to statistics
62H99 Multivariate analysis
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Bookstein, F. L.; Sampson, P. D.; Streissguth, A. P.; Barr, H. M., Exploiting redundant measurement of dose and developmental outcomenew methods from the behavioral teratology of alcohol, Dev. Psychol, 32, 3, 404-415 (1996)
[2] Elden, L., 2004. Partial least-squares vs. Lanczos bidiagonalization I: analysis of a projection method for multiple regression, Comput. Statist. Data Anal., in press; doi: 10.1016/S0167-9473(03)00138-5; Elden, L., 2004. Partial least-squares vs. Lanczos bidiagonalization I: analysis of a projection method for multiple regression, Comput. Statist. Data Anal., in press; doi: 10.1016/S0167-9473(03)00138-5 · Zbl 1429.62288
[3] Höskuldsson, A., PLS regression methods, J. Chemometrics, 2, 211-228 (1988)
[4] Höskuldsson, A., Causal and path modelling, Chemometrics Intell. Lab. Systems, 58, 2, 287-311 (2001)
[5] Kubberød, E.; Ueland, Ø.; Rødbotten, M.; Westad, F.; Risvik, E., Gender specific preferences and attitudes towards meat, Food Qual. Preference, 13, 5, 285-294 (2002)
[6] Kühn, B. F.; Thybo, A. K., The influence of sensory and physiochemical quality on Danish children’s preferences for apples, Food Qual. Preferences, 12, 534-550 (2001)
[7] Martens, H., Reliable and relevant modelling of real world dataa personal account of the development of PLS regression, Chemometrics Intell. Lab. Systems, 58, 85-95 (2001)
[8] Martens, H.; Martens, M., Multivariate Analysis of Quality. An Introduction (2001), Wiley: Wiley Chichester, UK
[9] Martens, H.; Naes, T., Multivariate Calibration (1989), Wiley: Wiley Chichester, UK · Zbl 0732.62109
[10] Nguyen, D.V., Rocke, D.M., 2004. On partial least squares dimension reduction for microarray-based classification: a simulation study. Comput. Statist. Data Anal., in press; doi: 10.1016/j.csda.2003.08.001; Nguyen, D.V., Rocke, D.M., 2004. On partial least squares dimension reduction for microarray-based classification: a simulation study. Comput. Statist. Data Anal., in press; doi: 10.1016/j.csda.2003.08.001 · Zbl 1429.62578
[11] Thybo, A. K.; Kühn, B. F.; Martens, H., Explaining Danish children’s preferences for apples using instrumental, sensory and demographic/behavioural data, Food Qual. Preference, 15, 53-63 (2003)
[12] Wold, H., 1982. Soft modelling. The basic design and some extensions. In: Jöreskog, K.G., Wold, H. (Eds.), Systems Under Indirect Observation. Causality-Structure-Prediction. Vol. I. North-Holland, Amsterdam, pp. 263-271.; Wold, H., 1982. Soft modelling. The basic design and some extensions. In: Jöreskog, K.G., Wold, H. (Eds.), Systems Under Indirect Observation. Causality-Structure-Prediction. Vol. I. North-Holland, Amsterdam, pp. 263-271.
[13] Wold, S., Martens, H., Wold, H., 1983. The multivariate calibration problem in chemistry solved by the PLS method. In: Ruhe, A., Kågström,B. (Eds.), Proceedings of the Conference on Matrix Pencils, March 1982, Lecture Notes in Mathematics, Springer, Heidelberg, pp. 286-293.; Wold, S., Martens, H., Wold, H., 1983. The multivariate calibration problem in chemistry solved by the PLS method. In: Ruhe, A., Kågström,B. (Eds.), Proceedings of the Conference on Matrix Pencils, March 1982, Lecture Notes in Mathematics, Springer, Heidelberg, pp. 286-293. · Zbl 0499.65065
[14] Wold, S., Hellberg, S., Lundstedt, T., Sjöström, M., Wold, H., 1987. PLS modeling with latent variables in two or more dimensions. In: Proceedings: PLS Model Building: Theory and Applications. Symposium Frankfurt am Main, September 23-25, 1987.; Wold, S., Hellberg, S., Lundstedt, T., Sjöström, M., Wold, H., 1987. PLS modeling with latent variables in two or more dimensions. In: Proceedings: PLS Model Building: Theory and Applications. Symposium Frankfurt am Main, September 23-25, 1987.
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.