GROEC: Combination Method via Generalized Rolling Origin Evaluation

Jose Augusto Fiorucci

Combination methods have performed well in time series forecast competitions. This study proposes a simple but general methodology to combine time series forecast methods. Weights are calculated using a cross-validation scheme that assigns greater weights to methods with more accurate in-sample predictions. The methodology was applied to combine forecasts from Theta, Exponential Smoothing, and ARIMA models, placing 5th in the M4 Competition for point and interval forecasting. (joint work with Francisco Louzada)