For combining fields of within a measurement the usage if pivot() is the certainly the way to go.
@Anaisdg wrote: Joins can be computationally expensive.
That’s what I found for in a case where one afaik really needs a join().
See posting Poor performance for join(): cpu and memory grows quadratically with row count .
So questions to @ThePie :
- did your original query with a
join()have decent performance ? - did you also see a quadratic dependence of row count ?
