Sequential Quantile Prediction of Time Series - Statistics > MethodologyReport as inadecuate

Sequential Quantile Prediction of Time Series - Statistics > Methodology - Download this document for free, or read online. Document in PDF available to download.

Abstract: Motivated by a broad range of potential applications, we address the quantileprediction problem of real-valued time series. We present a sequential quantileforecasting model based on the combination of a set of elementary nearestneighbor-type predictors called -experts- and show its consistency under aminimum of conditions. Our approach builds on the methodology developed inrecent years for prediction of individual sequences and exploits the quantilestructure as a minimizer of the so-called pinball loss function. We perform anin-depth analysis of real-world data sets and show that this nonparametricstrategy generally outperforms standard quantile prediction methods

Author: Gérard Biau LSTA, PMA, Benoît Patra LSTA



Related documents