Working paper

Prediction via the Quantile-Copula Conditional Density Estimator

Olivier Faugeras

Abstract

To make a prediction of a response variable from an explanatory one which takes into account features such as multimodality, a nonparametric approach based on an estimate of the conditional density is advocated and considered. In particular, we build point and interval predictors based on the quantile-copula estimator of the conditional density by Faugeras [8]. The consistency of these predictors is proved through a uniform consistency result of the conditional density estimator. Eventually, the practical implementation of these predictors is discussed. A simulation on a real data set illustrates the proposed methods.

Keywords

nonparametric estimation; modal regressor; level-set;

Replaced by

Olivier Faugeras, Prediction via the Quantile-Copula Conditional Density Estimator, Communications in Statistics - Theory and Methods, vol. 41, n. 1, 2012, pp. 16–33.

Reference

Olivier Faugeras, Prediction via the Quantile-Copula Conditional Density Estimator, TSE Working Paper, n. 09-124, December 7, 2009.

See also

Published in

TSE Working Paper, n. 09-124, December 7, 2009