Closed scottiegarcia closed 6 months ago
Mocking pd.DataFrame
in one of the tests was causing a new function in sklearn==1.4.1 to error. Since the erroring test_save_conf_mat
function is just trying to ensure that the correct confusion matrix is generated and other functions validate the f1_report, I figure just mocking that classification_report
function is the easiest path forward
This fix pertains to my recently created issue here. The JSON encoder was only expecting/being tested for
pd.Timestamp
as values for_dt_obj_min
and_dt_obj_max
, but I ran into a case where those fields weredatetime.datetime
objects.Running the newly created test on the lastest release reproduces the bug, running on this branch resolves it.