A hybrid model for GEFCom2014 probabilistic electricity price forecasting

This paper provides detailed information on Team Poland’s approach in the electricity price forecasting track of GEFCom2014. A new hybrid model is proposed, consisting of four major blocks: point forecasting, pre-filtering, quantile regression modeling and post-processing. This universal model structure enables independent development of a single block, without affecting performance of the remaining ones. The four-block model design in complemented by including expert judgements, which may be of great importance in periods of unusually high or low electricity demand.