Goupby may be the higher abstraction, but it's not necessarily better.
Dimensional models are basically modules (generalised linear spaces), a groupby can do the same things but doesn't really give much useful structure to work with (at best the result is ordet independent, most of the time).
This is also why sums and counts tend to be more useful than averages.
Dimensional models are basically modules (generalised linear spaces), a groupby can do the same things but doesn't really give much useful structure to work with (at best the result is ordet independent, most of the time).
This is also why sums and counts tend to be more useful than averages.