Ensembles of general regression neural networks for short-term electricity demand forecasting

Abstract

This work presents ensembles of general regression neural network for short-term electricity demand forecasting. Several types of ensembles are proposed which differ in the source of diversity of individual members. Diversity is generated using different subsets of training data, different subsets of features, randomly disrupted training data and randomly disrupted model parameters. Experimental study on several datasets demonstrates that ensemble learning leads to decreasing in forecast errors comparing to the mean errors of the base learners.