Beitrag in einem Sammelband
Sums of smooth exponentials
In: Bowman, A. (Ed.): Proceedings of the 25th International Workshop on Statistical Modelling, Glasgow 5 - 9 July, 2010, 113–118
Glasgow, UK, University of Glasgow, Department of Statistics (2010)
Abstract
Standard smoothing procedures are challenged if the smoothness of the curve to be fitted varies strongly over the domain. A model is presented that conceptualizes the structure of such data as a sum of smooth components, which are modelled on the log scale and are then additively combined. The resulting estimation algorithm leads to a penalized composite link model. The approach is illustrated by three data examples.