Forecasting with many predictors - Is boosting a viable alternative?

This paper evaluates the forecast performance of boosting, a variable selection device, and compares it with the forecast combination schemes and dynamic factor models presented in Stock and Watson (2006). Using the same data set and comparison methodology, we find that boosting is a serious competitor for forecasting US industrial production growth in the short run and that it performs best in the longer run.

MoreLess

Year of publication:	2010
Authors:	Buchen, Teresa ; Wohlrabe, Klaus
Publisher:	München : Ludwig-Maximilians-Universität München, Volkswirtschaftliche Fakultät
Subject:	Prognoseverfahren \| Faktorenanalyse \| Bayes-Statistik \| Theorie \| Industrieproduktion \| Konjunkturprognose \| USA \| Forecasting \| Boosting \| Cross-validation

More details

Series:	Munich Discussion Paper ; 2010-31
Type of publication:	Book / Working Paper
Type of publication (narrower categories):	Working Paper
Language:	English
Other identifiers:	10.5282/ubm/epub.11788 [DOI] 641666527 [GVK] hdl:10419/104321 [Handle] RePEc:lmu:muenec:11788 [RePEc]
Classification:	C53 - Forecasting and Other Model Applications ; E27 - Forecasting and Simulation
Source:	EconStor

Persistent link: https://ebvufind01.dmz1.zbw.eu/10010427583