Predicting Electricity Consumption with Random Walks on Gaussian Processes

2024/09/09 by Chloé Hashimoto-Cullen, Hashimoto-Cullen, Chloé, Benjamin Guedj +1
Computer Science · Engineering · #Applications (stat.AP) #Energy Load and Power Forecasting #FOS: Computer and information sciences #Gaussian Processes and Bayesian Inference #Machine Learning (cs.LG) #Methodology (stat.ME) #Smart Grid Energy Management

paper · pdf · doi:10.48550/arxiv.2409.05934

openalex publication_date 2024/09/09 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01

Abstract

We consider time-series forecasting problems where data is scarce, difficult to gather, or induces a prohibitive computational cost. As a first attempt, we focus on short-term electricity consumption in France, which is of strategic importance for energy suppliers and public stakeholders. The complexity of this problem and the many levels of geospatial granularity motivate the use of an ensemble of Gaussian Processes (GPs). Whilst GPs are remarkable predictors, they are computationally expensive to train, which calls for a frugal few-shot learning approach. By taking into account performance on GPs trained on a dataset and designing a random walk on these, we mitigate the training cost of our entire Bayesian decision-making procedure. We introduce our algorithm called Domino (ranDOM walk on gaussIaN prOcesses) and present numerical experiments to support its merits.

Related

• Daily Middle-Term Probabilistic Forecasting of Power Consumption in North-East England
• Energy consumption forecasting using a stacked nonparametric Bayesian\n approach
• A Machine Learning approach to Risk Minimisation in Electricity Markets\n with Coregionalized Sparse Gaussian Processes
• Online Hierarchical Forecasting for Power Consumption Data
• Conditional Approximate Normalizing Flows for Joint Multi-Step Probabilistic Forecasting with Application to Electricity Demand