Abstract
In this article, we consider the estimation of the regression function in a dependent biased model. It is assumed that the observations form a stationary α-mixing sequence. We introduce a new estimator based on a wavelet basis. We explore its asymptotic performances via the supremum norm error and the mean integrated squared error. Fast rates of convergence are established.
Keywords:
AMS Subject Classifications :
Acknowledgements
The authors thank the referees and the associated editor for insightful comments that helped them to improve the article significantly.