From Distributional to Quantile Neural Basis Models: the case of Electricity Price Forecasting
This work addresses the need for interpretable probabilistic forecasting in electricity price prediction, though it is incremental as it builds on existing neural and quantile regression methods.
The paper tackled the challenge of interpretability in neural network-based probabilistic forecasting by introducing the Quantile Neural Basis Model, which integrates interpretability principles from Quantile Generalized Additive Models into an end-to-end neural framework, achieving predictive performance comparable to existing distributional and quantile regression neural networks on day-ahead electricity price forecasting.
While neural networks are achieving high predictive accuracy in multi-horizon probabilistic forecasting, understanding the underlying mechanisms that lead to feature-conditioned outputs remains a significant challenge for forecasters. In this work, we take a further step toward addressing this critical issue by introducing the Quantile Neural Basis Model, which incorporates the interpretability principles of Quantile Generalized Additive Models into an end-to-end neural network training framework. To this end, we leverage shared basis decomposition and weight factorization, complementing Neural Models for Location, Scale, and Shape by avoiding any parametric distributional assumptions. We validate our approach on day-ahead electricity price forecasting, achieving predictive performance comparable to distributional and quantile regression neural networks, while offering valuable insights into model behavior through the learned nonlinear mappings from input features to output predictions across the horizon.