Selection of a NLR-Model by Re-Sampling Technique

The aim of this study is to discuss a method for the selection of parametric models of the moisture retention characteristic MRC for natural soils. The author presents computational tools in S-Plus for the adequate selection of these parametric models as non-linear regression models (NLR-models) and this method of selection based on cross-validation and bootstrap (re-sampling technique). Early, these current computational tools were developed in some works about the selection of the nonlinear regression models. The current results will be compared with the obtained results which are based on the simulation studies and some calculations of measures of nonlinearity in NLR models. Finally, the results indicate to the best selected model which is identical with the selected model by the simulation study and the calculation of the measures of nonlinearity.