Prévision de la demande électrique journalière d'une maison à l'aide de modèles de séries temporelles et de Machine Learning

Abstract

This Master's end-of-study project focuses on forecasting residential electricity consumption using Machine Learning techniques. The primary objective is to develop predictive models capable of accurately estimating electricity demand based on historical consumption data and relevant explanatory variables. Following a comprehensive literature review, four prediction models—Linear Regression, Random Forest, XGBoost, and CatBoost—were developed, implemented, and comparatively evaluated. The performance of these models was assessed using Mean Absolute Error (MAE), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and the coefficient of determination (R2). The experimental results demonstrate the effectiveness of machine learning techniques in improving forecasting accuracy. The best-performing model achieved an MSE of **0.408103**, an RMSE of **0.638829**, an MAE of **0.480083**, and an **R2 score of 0.986685**, indicating excellent predictive performance. These findings highlight the potential of machine learning to support efficient energy management, accurate electricity demand forecasting, and improved electrical grid operation.