Multilayer perceptron for GEFCom2014 probabilistic electricity price forecasting

This paper proposes a forecasting approach based on a feedforward neural network for probabilistic electricity price forecasting for GEFCom2014. The approach does not require any special data preprocessing, such as detrending, deseasonality or decomposition of the time series. The input variables, zonal and system loads are processed nonlinearly by the multilayer perceptron in order to obtain point forecasts. The model is trained on the most recent period of data. This allows us to take into account the current trends, conditions and variability of the processes, as well as to simplify the model. Probabilistic forecasts are then generated from the point forecasts and the error distribution on the training set in the form of quantiles.