Who it's for
Platform
Resources
subtopic
Compute two easy baselines on the test period: 'same hour last week' (seasonal naive) and your LightGBM model from the ML module. Any neural network must beat the first and come close to the second to be worth it.
Builds on: M03 路 Build: Model-Ready Data & Baselines, S08 路 Build: Boosting (the tabular state of the art)