An evaluation of the PoLiSh smoothed regression and the Monte Carlo Cross-Validation for the determination of the complexity of a PLS model

A crucial point of the PLS algorithm is the selection of the right number of factors or components (i.e., the determination of the optimal complexity of the system to avoid overfitting). The leave-one-out cross-validation is usually used to determine the optimal complexity of a PLS model, but in practice, it is found that often too many components are retained with this method. In this study, the Monte Carlo Cross-Validation (MCCV) and the PoLiSh smoothed regression are used and compared with the better known adjusted Woldʹs R criterion.