Comparative Analysis of Electricity Market Prices Based on Different Forecasting Methods

Abstract

With the increasing share of renewable energy in the energy markets, the wholesale prices of electricity tend to decrease due to the merit order effect, thus creating a need to predict the future prices of electricity to avoid economical losses and to maximize profits. The goal of this paper is to forecast day-ahead electricity prices for Germany, based on different input parameters, and choosing the best fitting model. Forecasts are built using ARIMA, SARIMA, SARIMAX, and multiple linear regression methods. It is observed that the SARIMAX model performed best, followed by SARIMA, multiple linear regression, and ARIMA. The introduction of exogenous variables in the SARIMAX model exhibited significant effects on the electricity price forecasting, with the lowest errors and highest correlation factor.