Causal Feature Selection for Weather-Driven Residential Load Forecasting

TL;DR

This paper explores the use of causal feature selection to improve short-term residential load forecasting by identifying the most relevant weather variables, leading to more robust and parsimonious models.

Contribution

It demonstrates that causal feature selection can effectively identify key weather drivers, enhancing model simplicity and robustness compared to non-causal methods.

Findings

Causal selection emphasizes direct thermal drivers over correlational variables.

Causal feature selection improves robustness during extreme weather conditions.

The approach offers a practical method for integrating weather data into load forecasting.

Abstract

Weather is a dominant external driver of residential electricity demand, but adding many meteorological covariates can inflate model complexity and may even impair accuracy. Selecting appropriate exogenous features is non-trivial and calls for a principled selection framework, given the direct operational implications for day-to-day planning and reliability. This work investigates whether causal feature selection can retain the most informative weather drivers while improving parsimony and robustness for short-term load forecasting. We present a case study on Southern Ontario with two open-source datasets: (i) IESO hourly electricity consumption by Forward Sortation Areas; (ii) ERA5 weather reanalysis data. We compare different feature selection regimes (no feature selection, non-causal selection, PCMCI-causal selection) on city-level forecasting with three different time series forecasting models: GRU, TCN, PatchTST. In the feature analysis, non-causal selection prioritizes radiation and moisture variables that show correlational dependence, whereas PCMCI-causal selection emphasizes more direct thermal drivers and prunes the indirect covariates. We detail the evaluation pipeline and report diagnostics on prediction accuracy and extreme-weather robustness, positioning causal feature selection as a practical complement to modern forecasters when integrating weather into residential load forecasting.