Closed fsaad closed 6 years ago
For the test FAIL: distributions.tests.test_models.test_add_remove('lp.models.niw',)
it seems that other models do not return an exact match for the marginal equaling the product of conditions. I'm not sure on what basis the threshold are declared, or whether it makes sense to either report an xfail or adjust the threshold slightly.
Looks like a nuisance random failure to me. I'd recommend either twiddling a random number seed or slightly increasing a threshold.
It seems that lp.models.niw
is already known to be buggy, should have checked these issues:
https://github.com/posterior/distributions/issues/6 https://github.com/posterior/loom/issues/4
PR #16 addresses the numpy error.
Using
make all
results in two failures in the distributions test suite:The numpy error in
distributions.tests.test_util.test_scores_to_probs
is due to deprecation: https://github.com/numpy/numpy/issues/6464Not sure why the probabilities are failing.