Smoothing Techniques: With Implementation in S
Springer-Verlag Berlin and Heidelberg GmbH & Co. K
W. Hardle (Catholic University of Louvain, France)
The author has attempted to present a book that provides a non-technical introduction into the area of nonparametric density and regression function estimation. The application of these methods is discussed in terms of the S computing environment. Smoothing in high dimensions faces the problem of data sparseness. A principal feature of smoothing, the averaging of data points in a prescribed neighbourhood, is not really practicable in dimensions greater than three. Additive models provide a way out of this dilemma, but they require highly effective algorithms. For this purpose, the method of WARPing (Weighted Averaging using Rounded Points) is described in detail in the text.