@boaguilar I removed the N > 0 check for the t_test_paired, as this was consistent with how we'd previously implemented the t_test (students) UDF. I don't have any opinion one way or another — if the check is present, the query will just return zero results; otherwise, it'll return a struct with the values shown below (including NaN).
I'm not really sure what's more "correct", though maybe easier for the user to handle the NaN case since the result always has the same shape?
@boaguilar I removed the
N > 0
check for thet_test_paired
, as this was consistent with how we'd previously implemented thet_test
(students) UDF. I don't have any opinion one way or another — if the check is present, the query will just return zero results; otherwise, it'll return a struct with the values shown below (including NaN).I'm not really sure what's more "correct", though maybe easier for the user to handle the NaN case since the result always has the same shape?