There is an important question about treatment of federalized states where data may not be collected by national government (but is collected locally) or situations where underlying service does not exist (e.g. no national transport like rail). In such cases it has been discussed whether “n/a” should be the answer (with appropriate implications for scoring).
There was extensive discussion of this in 2013, see e.g. here:
I would distinguish the case of:
A. Dataset simply does not exist / not relevant (e.g. no national train system)
B. Dataset isn’t very important (national train system is marginal)
C. Infrastructure does exist but is run federally (or privately)
For my part, I’m still not convinced that N/A is right answer except in A which is very rare.
On C which is most common afaict I think you can still couldn’t have national provision of this dataset even if just aggregating or pointing to local data. As a concrete example, UK rail is broken up amongst many different private companies yet government has still made effort to get train timetables made option.