ABSTRACT
In subsequent chapters we will study several families of estimators for the regression function. Each family is indexed by a parameter (often called the smoothing parameter). The selection of an estimator from a given family requires the choice of a value for this index parameter. While such a choice can be made subjectively, an objective choice will usually be preferred, at least as a starting point for subsequent estimator fine tuning. The present chapter provides an overview of several techniques which have been found useful for smoothing parameter selection. Their usefulness is not restricted to nonparametric regression problems and they can, in fact, be used to solve a variety of estimation problems, some of which will be explored in examples and exercises.
