Fig. 9

Effect of regularisation for all the case studies. Violin plots illustrating the effect of regularisation on the fitting and cross-validation errors (given as normalized root mean square error, NRMSE) for each model and over all the data sets considered. It is shown how regularisation increases the NRMSE for the fitting, but with the benefit of generally improving the predictive power, i.e. reducing the NRMSE in the cross-validations