Prediction of California Bearing Ratio from Index Properties of Soils Using Parametric and Non-parametric Models

This work proposes a methodology to obtain from the soils properties the best prediction model for the California bearing ratio index. The methodology proposes three different prediction techniques: (1) the multiple linear regression, a classical parametric technique; and two non-parametric techniques: (2) the local polynomial regression (LPR) and (3) the radial basis network. The LPR is a known statistical method, but in the geotechnical engineering field is not in common use. Besides, although several research works have been published in this field, they do not include a robust procedure for making good comparison between different models. Here, a cross validation method is proposed with this aim. A data set of 96 samples from Peruvian soils is used to illustrate the methodology. To validate the proposed methodology, a data set from the literature is also analyzed.