Functional estimation

This article presents the non-parametric functional estimation methods. These parametric model present, in general, a parameter of interest in the infinite dimension; most often this parameter is a function that one tries to estimate. It particularly focuses on the density by projection , distribution function and spectral density methods. These methods are of great interest being resistant to changes in models. They also allow for assisting statisticians in choosing a parametric model and are very efficient for forecasting. This presentation is illustrated by several applications.