Combinación de pronósticos usando una neurona multiplicativa simple generalizada (Forecast combining using a generalized single multiplicative neuron)

Juan D. Velásquez (jdvelasq@unal.edu.co), Cristian O. Zambrano (cozambra@unal.edu.co), Carlos J. Franco (cjfranco@unal.edu.co)


Universidad Nacional de Colombia
This paper appears in: Revista IEEE América Latina

Publication Date: June 2014
Volume: 12,   Issue: 4 
ISSN: 1548-0992


Abstract:
Forecast combining is an important technique for increasing the accuracy of the forecasts obtained using different alternative models. In this article, we propose the use of a generalized single multiplicative neuron as a nonlinear combiner inside of a forecasts combination model. Numerical evidences indicate that, at least for the experimental case, the multiplicative neuron is able to obtain forecasts more accurate that each individual model and the forecasts combination using a simple arithmetical average.

Index Terms:
Combination of forecasts, ensemble methods, Holt-Winters model, neural networks, SARIMA   


Documents that cite this document
This function is not implemented yet.


[PDF Full-Text (266)]