We argue that the future of the most skilful weather forecasts will likely come from machine learning stacking, by the very nature that stacking typically yields performance beyond than any base model
Almost like WeatherUnderground meets AI. Loved the SHAP value chart.
This is really interesting work! Did you experiment with any other methods besides the XGBoost Regressor to "stack" the model outputs?
Almost like WeatherUnderground meets AI. Loved the SHAP value chart.
This is really interesting work! Did you experiment with any other methods besides the XGBoost Regressor to "stack" the model outputs?