Name of QuantLet: Redispatch_ML_Forecasting

Published in: Day-ahead Forecasting for Redispatch Measures using Machine Learning Models

Description: Day-ahead forecasting of realized German TSO redispatch load with classical, tree-based and deep-learning models (NBEATSx, NHiTS, TFT). Includes the rolling-window evaluation, Model Confidence Set, F-beta event detection, circular-shift covariate ablation and Integrated-Gradients feature importance, all reproduced from the original data via the paper_results_cli module.

Keywords: redispatch, congestion management, day-ahead forecasting, deep learning, NBEATSx, NHiTS, temporal fusion transformer, model confidence set, integrated gradients, ENTSO-E

Author: Vlad Bolovaneanu

Submitted: Monday, 10 August 2026