Forecast plugin: Losing columns

lisa811

Hi,

I am using the new forecast plugin. When applying to my dataset, the only columns kept are the date and the target. So I am losing my id column, which I need to join back to another dataset.

There would be two possibilities: My datasets are partitioned by "id_1", so if it is somehow possible to convert the partitions to be displayed in a column it would help. On the other hand, I have a column "id" containing the same values as "id_1" in my input dataset for the forecast plugin. So if it would be possible to keep this column it would also be fine.

Is there any solution to that?

Thanks and best regards,

Lisa

StanG

Hi,

If I understand well, you have a column with a constant value that you want to retrieve after the forecast.
To do so, you can select long format in the train and evaluate recipe and make this constant value column a time series identifier. This way this column will be retrieved.

You can also retrieve informations about the partitions using this processor in a prepare recipe: https://doc.dataiku.com/dss/8.0/preparation/processors/enrich-with-record-context.html

lisa811

Hey,

Great, thanks a lot! I did not know about this processor, this solves my problem!