Derived timeseries of "Categorical time series" (strings)

Ask the Community
1

I have multiple time-series that are Categorical, aka, they are not integer or floats, but strings types.
I would like to have a derived timeseries that acts as a “fallback” between different timeseries, for instance:
If time-series-A is present, take time-series A. Else, take time-series-B. Else, take time-series-C.

I want to use the Combine time series card.

image
image3068×1002 139 KB

It seems that in the derived time-series UI, an error occur if I’m trying to derived new categorical time-series with A node has an input with an incompatible type. It's likely a categorical series was used as a numeric input, or vice versa.

Is there a way to achieve this ?

2

I think in this case the combine time series card might not support categorical series which is why you’re getting that error. I think you can maybe try to hack this with a time series formula (i.e. use if/else statements in the formula to determine which formula’s values to use).

Edit: This was actually an issue with the implementation of the combine time series card that will be fixed soon but this should work.