You may want to blend with
the two distinct Secondary datasources --
each one on either [City] or [State].
The results may be looking counter-intuitive (or even weird).
For example when 'rolling-up' (removing) [City] --
and leaving only the [State] dimension on a view --
the calculation [YF : City Marker] gives 1.11 for the Colorado
(which is likely to come from the [City] == 'Denver').
This could bring the confusion later --
when referring the field in other calcs.
To make it a bit simple, one could be using joins instead.
Joining the 'lookup table' twice -- ON City and ON State --
would allow the calculation to be a Row-Level.
Please find the attached.
Data Blend Sample_v10.2.twbx 2.4 MB