Parameterize the columns for their forecast usage: series, time, target, cov_attribute, cov_unavailable, cov_available
Adjust AutoML Forecasting component to use the covariates
Handle missing data in the data preparation steps
Create a component to create an champion ensemble across all methods tried by using Vizier to create a optimal weighting (minimize MAPE, MAE, pMAE) across the test region
Update champion methods to create a combined table ready for reporting for the champion method: combines original data, test region, horizon region
Update diagnostic plots for data preparation and results review
This Branch updates the 04g Notebook: