Review Stephan Rasp
I thank the authors for their clear description and code for their submission. The authors apply a modified version of a method they had previously published. The method is based on a Bayesian framework, only considering the raw forecast of the variable to be predicted as input. The train/valid split is correct and no overfitting is to be suspected. Therefore, I give this entry a pass.
I have one additional questions for the authors:
- In the original paper you also used large scale climate indices as predictors but I don’t think that was used for this competition. What was the reason for this?
Best, Stephan Rasp