, Volume 17, Issue 1, pp 89-102

Local roughness penalties for regression splines


This paper introduces a new nonparametric estimator of the regression based on penalized regression splines. Local roughness penalties that rely on local support properties of B-splines are introduced in order to deal with the spatial heterogeneity of the function to be estimated. This estimator is shown to attain optimal rates of convergence. Then its good performances are confirmed on a simulation study.