ABSTRACT
In this article, we consider an alternative estimator of the conditional mode when the explanatory variable takes values in a semimetric space. This alternative estimate is based in a recursive kernel method. Under the ergodicity hypothesis, we quantify the asymptotic properties of this estimate, by giving the almost complete convergence rate. The asymptotic normality of this estimate is also given. Our approach is illustrated by a real data application.
Acknowledgments
The authors thank the associate editor and the anonymous reviewer for their valuable comments and suggestions which improved substantially the quality of an earlier version of this article.
Notes
1 Let be a sequence of real r.v.’s; we say that
converges almost completely (a.co.) to zero if, and only if,
,
. Moreover, we say that the rate of almost complete convergence of
to zero is of order
(with
and we write
if, and only if,
,
. This kind of convergence implies both almost sure convergence and convergence in probability.