The functional parameter object used in the beta basis system is then redefined based on the refined smoothing parameter identified in order to ensure that the functional linear regression analysis converges on a model that has the best chances of performing well out-of-sample.