winedarksea
commented
1 year ago I am not quite sure I follow, but you have a few ways of passing in data like this.
future_regressor handles data you will definitely know about the future. A covariate without known future can be reshaped like this with create_regressor
You can pass in covariates as multivariate data, and some models, multivariate models, will handle them.
It is definitely a bit confusing
How can I include past and future covariates in autots? I like to use a historical training set and then apply it to different segments in the future. For example training stops on Sunday and forecast starts either Monday, Tuesday, Wednesday... Thank you