Day-ahead electricity consumption prediction of a population of households: analyzing different machine learning techniques based on real data from RTE in France

Abstract

Forecasting of power consumption has been a topic of great interest for the stakeholders of electricity markets. It has an essential role in decision making, including purchasing and generating electric power, load switching, and demand side management. Different algorithms are tested and used for balancing the demand and supply of energy. This research work focuses on predicting power consumption using time series forecasting methods for the Île-de-France region with publicly available energy data from RTE, France. The two machine learning algorithms Support Vector Machine (SVM) and Recurrent Neural Network (RNN) are implemented and tested for their accuracy in predicting day-ahead half-hourly power consumption data. This paper provides brief insights on the algorithms used and further explains the data handling for its implementation. The Mean Absolute Percentage Error (MAPE) is used as the performance measure. The results indicate a higher accuracy of the RNN at the cost of longer computation times.