Graph neural networks show limited edge benefits in load forecasting
A Diagnostic Protocol for GNNs in Electricity Load Forecasting: Separating Parameter Sharing from Neighborhood Access
Artificial Intelligence
Summary
Forecasting electricity demand often involves looking at data from multiple locations. The authors studied graph neural networks (GNNs), which can share information across locations and also learn from neighboring sites. Their work separates the benefit of sharing model settings from the benefit of using neighbor data. They found that much of the improvement in forecasting comes from sharing settings rather than neighbor connections, especially when using short recent data. This helps decide when using a graph structure is truly beneficial.
What this means in practice
- •For energy forecasting teams: Improve forecasting models by distinguishing whether sharing parameters or neighbor data drives accuracy gains in electricity load predictions.
- •For data scientists in retail energy: Design better models for short-term demand forecasting by evaluating the true value of graph edges among client consumption series.
Authors
Eloi Campagne, Yvenn Amara-Ouali, Yannig Goude, Itai Zehavi, Argyris Kalogeratos
Abstract
A graph neural network (GNN) forecasting a collection of load series combines two mechanisms, sharing parameters across series and reading neighboring series, and a gain over a non-graph model may come from either; telling them apart decides whether a graph is worth maintaining. We propose a model diagnostic protocol that compares the GNN with three reference models: a per-node network, a network shared across nodes without neighbors, and the same GNN without edges. Degree-preserving rewiring tests whether neighbor identity matters. After a tuned benchmark on regional load in France and on low-voltage feeders in the UK, we apply the protocol to deliberately short node histories, as when non-stationary demand leaves only recent observations relevant: the feeders, within one week, and a public dataset of client consumption series. On the feeders, sharing parameters recovers most of the error gap between node-specific models and GNNs, about two thirds against a regularized one. Adding the edges to an otherwise identical backbone brings a smaller gain, half or more of which disappears under rewiring. On the client series, the neighbor mean degrades the per-node model on the shorter window and the edges' gain is not consistently distinguishable from zero; on the regions of France, where each node's own history is already informative, the ranking between per-node models and GNNs changes with tuning. A synthetic study points to noisy local measurements as one regime where neighbors help. In an exploratory online aggregation, graph and non-graph forecasters appear complementary, and the weight assigned to GNNs is not proportional to their contribution. In none of the three datasets do the edges account for most of a GNN's advantage over node-specific models, so a forecasting gain should be credited to the graph only after comparison with per-node and shared references.