Human spatial dynamics for electricity demand forecasting: the case of France during the 2022 energy crisis

Abstract

Accurate electricity demand forecasting is crucial to meet energy security and efficiency, especially when relying on intermittent renewable energy sources. Recently, massive savings have been observed in Europe, following an unprecedented global energy crisis. However, assessing the impact of such crisis and of government incentives on electricity consumption behaviour is challenging. Moreover, standard statistical models based on meteorological and calendar data have difficulty adapting to such brutal changes. Here, we show that mobility indices based on mobile network data significantly improve the performance of the state-of-the-art models in electricity demand forecasting during the sobriety period. We start by documenting the drop in the French electricity consumption during the winter of 2022-2023. We then show how our mobile network data captures work dynamics and how adding these mobility indices outperforms the state-of-the-art during this atypical period. Our results characterise the effect of work behaviours on the electricity demand.

References in corpus (7)

• Variable selection in nonparametric additive models
• Understanding vehicular routing behavior with location-based service data
• Using Mobility for Electrical Load Forecasting During the COVID-19 Pandemic
• A Survey on Principles, Models and Methods for Learning from Irregularly Sampled Time Series
• SHAFF: Fast and consistent SHApley eFfect estimates via random Forests
• Naive imputation implicitly regularizes high-dimensional linear models
• Bridging AI and Energy Forecasting: An Autonomous Workflow with Customized Toolkit