Predicting Electricity Consumption with Random Walks on Gaussian Processes

TL;DR

This paper introduces Domino, a novel algorithm that uses random walks on Gaussian Processes to efficiently forecast short-term electricity consumption in France, addressing data scarcity and computational challenges.

Contribution

We propose Domino, a new method leveraging random walks on Gaussian Processes to reduce training costs in short-term electricity consumption forecasting.

Findings

Domino outperforms traditional GPs in computational efficiency.

The method achieves comparable accuracy with fewer training samples.

Numerical experiments validate the effectiveness of Domino.

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 \textsc{Domino} (ranDOM walk on gaussIaN prOcesses) and present numerical experiments to support its merits.