In smart buildings, load forecasting models are often trained offline on dense, multivariate, high-frequency data, but only hourly, feature-limited inputs may be available during deployment. This necessitates the reconstruction of missing features, which can propagate errors through the model. If this input uncertainty is not accounted for, prediction intervals may become miscalibrated, affecting demand-response scheduling.
We investigate where uncertainty should be placed after reconstructing inference inputs and develop a unified one-day-ahead probabilistic forecasting framework that aligns temporal resolution, reconstructs unavailable inputs, and derives causal features. We compare a modular post-hoc residual-quantile scheme with an integrated in-model quantile-learning scheme using three mid-scale Deep Learning (DL) backbones: recurrent, hybrid recurrent, and attention-based Temporal Fusion Transformer (TFT) models, under identical inputs, forecasting horizons, preprocessing rules, and training budgets.
Results show that uncertainty placement is backbone-dependent. Integrated quantile learning is most reliable with the TFT, yielding 2.2-3.6% Mean Absolute Percentage Error (MAPE) and 28-83W Root Mean Square Error (RMSE) on the labeled test window, while producing intervals about 5x narrower than the modular intervals at the closest-to-nominal coverage level. Diebold-Mariano tests support the TFT ranking and the mixed behavior of the recurrent backbones. A reconstruction-sensitivity test shows that reconstructed inputs increase the Quantile Score (QS) by 106% while interval width remains nearly unchanged, indicating that the model does not automatically absorb reconstruction-induced uncertainty. Robustness checks against non-DL baselines and seasonal hold-out weeks support this ranking. Our results expose the limits of post-hoc residual quantiles when inference depends on reconstructed inputs.
Blogger's Review: This paper delves into the uncertainty issues in load forecasting for smart buildings, proposing an innovative framework to handle uncertainties arising from input reconstruction. By comparing the performance of different models, the authors demonstrate the advantages of integrated learning in addressing complex forecasting tasks, providing significant insights for future research.