Abstract
This paper presents a hybrid model for electricity price forecasting with focus on price spikes predictions. Nowadays, short-term forecasts have become increasingly important since the rise of the competitive spot electricity markets. A two-layered model is introduced for forecasting 7-days ahead hourly electricity price values of electricity spot market. Due to the importance of improved analysis of spikes for risk management, price segmentation into normal range and price spike module is applied. Price spike module consists of two segments: obtaining the probability of price spike occurrence and predicting the value of price spike. To avoid reliance on a single classifier, the compound classifier is proposed in the paper, which combines three individual classification methods: a support vector machine (SVM) classification, decision trees (DT) and probabilistic artificial neural network (PANN). The k-nearest neighbors algorithm (k-NN) is applied for the price spike value prediction.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call