Chapter 12 - Regression and conditional regression quantiles

Abstract

Probably, the most used technique in nonparametric estimation is nonparametric Nadaraya–Watson regression estimation. As a simplification, to avoid the curse of dimensionality, additive models can be used. It is possible to use a local Gaussian approach to nonparametric regression estimation, but it is much less direct than the Nadaraya–Watson estimator. It still t can be utilized, and the pairwise simplification introduced in earlier chapters may work in cases where additive models fail. More work is required to assess the true potential of the local Gaussian approach. In this chapter, we obtain some preliminary results and outline possible ideas, including treating conditional regression quantiles. We also give a simple simulated example.