Failing test: tests/clv/models/test_shifted_beta_geo.py::TestShiftedBetaGeoModel::test_distribution_customer_churn_time

[juanitorduz]

I am seeing the test tests/clv/models/test_shifted_beta_geo.py::TestShiftedBetaGeoModel::test_distribution_customer_churn_time fail in the CI (see for example https://github.com/pymc-labs/pymc-marketing/actions/runs/8606924039/job/23586352229?pr=619, but it happens in all recent PRs) with

args = (<function assert_allclose.<locals>.compare at 0x7f6033ec4430>, array([7.84 , 8.647, 8.493]), array([10.        ,  2.        ,  1.11111111]))
kwds = {'equal_nan': True, 'err_msg': '', 'header': 'Not equal to tolerance rtol=0.05, atol=0', 'verbose': True}

    @wraps(func)
    def inner(*args, **kwds):
        with self._recreate_cm():
>           return func(*args, **kwds)
E           AssertionError: 
E           Not equal to tolerance rtol=0.05, atol=0
E           
E           Mismatched elements: 3 / 3 (100%)
E           Max absolute difference: 7.38188889
E           Max relative difference: 6.6437
E            x: array([7.84 , 8.647, 8.493])
E            y: array([10.      ,  2.      ,  1.111111])

/opt/hostedtoolcache/Python/3.10.14/x64/lib/python3.10/contextlib.py:79: AssertionError

but I can not reproduce locally with pymc 5.12 ... could it be a 5.13 issue ? FYI @ColtAllen

[wd60622]

Seeing this failure but also experiencing a change in one of test outcomes for mocking prior

[juanitorduz]

I can confirm in pymc 5.12 it passes but fails on pymc 5.13

[juanitorduz]

With pymc 5.12 the test passes and the values being compared are

(Pdb) res.mean(("chain", "draw"))
<xarray.DataArray 'churn' (customer_id: 3)> Size: 24B
array([9.857, 2.012, 1.13 ])
Coordinates:
  * customer_id  (customer_id) int64 24B 0 1 2
(Pdb) stats.geom(customer_thetas).mean()
array([10.        ,  2.        ,  1.11111111])

[wd60622]

What is the latest on this?

[juanitorduz]

I have not been able to fix it. I'll try in the upcoming days. Maybe @ricardoV94 can take a look sometime in the next days 😄