Robust nonparametric regression estimation (Q1263900)

Following \textit{P. J. Huber}'s [Ann. Math. Statistics 35, 73-101 (1964; Zbl 0136.398)] proposal for location, the authors define a robust conditional location functional in order to obtain from it robust and nonparametric estimators of the conditional expectation. The functional is defined without requiring any moment condition. The authors give a method for obtaining consistency and strong convergence rates of robust nonparametric estimates from corresponding results for conditional empirical distribution functions. They obtain strongly consistent robust estimators of the conditional expectation by applying the functional equation to the empirical distributions considered by \textit{C. J. Stone} [Ann. Stat. 5, 595-645 (1977; Zbl 0366.62051)] and they give some examples by using nearest neighbor weights or weights based on kernel methods. Consistency results are obtained without requiring any regularity condition on the distribution of (X,Y). Strong convergence rates and the asymptotic distribution of the proposed estimators are derived.