I have a very simple question but I cannot seem to find a reasonable solution.
I am using Azure automl to predict time series. So far so good.
As you can see in the image below, I would like to train a model and then later on run it by giving it more recent context. In other words, I would like to train a model from January 2022 to December 2022 and then use it to predict February 2023.
My issue is, I would like to give it context information that is also present in training data. I think this makes perfect sense, so, for example, if I want to predict February 2023 I would like to give it context from January 2023 December 2022, November 2022 etc.
This seems to be impossible. Any overlap between training and context data will give this error:
"Input prediction data X_pred or input forecast_destination contains dates prior to the latest date in the training data. Please remove prediction rows with datetimes in the training date range or adjust the forecast_destination date".
Isn't this a big limitation? What am I missing?
