An evaluation of forecasting methods and forecast combination methods in goods management systems

Carsten Schneider, Matthias Klapper and Thomas Wenzel

In this paper we use 4 different time series models to forecast sales in a goods management system. We use a variety of forecast combining techniques and measure the forecast quality by applying symmetric and asymmetric forecast quality measures. Simple, rank-, and criteria-based combining methods lead to an improvement of the individual time series models.

MoreLess

Year of publication:	1999
Authors:	Schneider, Carsten ; Klapper, Matthias ; Wenzel, Thomas
Publisher:	Dortmund : SFB 475, Universität Dortmund
Subject:	ARX forecasts \| dynamic linear model \| combination of forecasts \| goods management system \| asymmetric loss function \| Marktforschung (STW) \| Prognoseverfahren (STW) \| Theorie (STW) \| Marktforschung \| Market research \| Prognoseverfahren \| Forecasting model \| Theorie \| Theory