Title: Combining temporal disaggregation forecasts with artificial neural networks

Authors: Leila Hedhili Zaier

Addresses: High School of Business, University of Tunis, Tunisia; Institut Supérieur de Gestion, 41, Rue de la liberté – Cité Bouchoucha, Le Bardo 2000-Tunisie, Tunisia

Abstract: This paper investigates the use of artificial neural networks (ANNs) to combine time series forecast of high frequency data obtained from various temporal disaggregation methods without related series (e.g., Almon, 1988; Lisman and Sandee, 1964; Boot and Feibes, 1967; Boot et al., 1967; Stram and Wei, 1986; Zaier and Abed, 2014). We use the example of deriving quarterly US GDP from annual one to evaluate the performance of the proposed method. We demonstrate that combining with nonlinear ANNs generally produces better forecasts than forecasts obtained from individual temporal disaggregation methods and from also traditional linear combining procedures on the basis of performance measures.

Keywords: temporal disaggregation; time series forecasts; high frequency data; artificial neural networks; ANNs; forecasting.

DOI: 10.1504/IJBIDM.2014.068457

International Journal of Business Intelligence and Data Mining, 2014 Vol.9 No.4, pp.318 - 329

Published online: 10 Apr 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article