Heterogeneous Ensembles for Short-Term Electricity Demand Forecasting

In this work multi-model ensembles are proposed for short-term electricity demand forecasting. The ensembles are composed of ten members representing different model classes. The base models are integrated using simple averaging or dynamically weighted averaging, where weights depend on the model performance on the forecasting tasks similar to the current one. Simulation studies on different load time series show that the ensemble errors are lower than the mean errors for the base models, and in most cases even lower than error of the best base model. The variance of the forecasts is also reduced.