Training Program
Module 3: PLS regression methods & extended tools
- Introduction: regression and multicollinearity problems, p> n, missing values.
- Adjustment: dimensionality reduction (PCR, PLS), penalty (Ridge, Lasso, Elastic net), regulation & bias,
- Power Method, NIPALS, PCA and missing values, NIPALS geometry, sparse PCA.
- PLS1-NIPALS, PLS2-NIPALS, SIMPLS
- Model validation, choice of the number of components, bootstrap for coefficients, information criterion AIC / BIC number of degrees of freedom,
- PLS1 and Lanczos methods, PLS1 and Krylov subspaces.
- PLS1 Geometry
- PLS1 and optimization of a criterion: Tucker's criterion
- PLS and variable selection: Sparse PLS
- OPLS
- Discriminant PLS: PLS-DA, Barker & Rayen approach
- PLS and very large dimentions: Kernel PLS, nonlinear PLS.
- Generalized PLS: Logistic-PLS, PLS-Poisson, PLS-Cox, PLSDR
- Applications, Softs
CHIMIOMETRIE 2010 Symposium
2010 Challenge Winners...
Vous êtes le ième visiteurs
Dernière mise à jour : 07/10/2016