[MRG] DOC covariance doctest examples #12124
Conversation
|
Removed the elliptic example, it shows the same issue as what I saw in PR #11732 |
There was a problem hiding this comment.
Very nice, thanks @adrinjalali ! Actually the cov.location_ is not documented in the docstring, maybe add it as well?
| >>> np.random.seed(0) | ||
| >>> X = np.random.multivariate_normal(mean=[0, 0], | ||
| ... cov=real_cov, | ||
| ... size=300) |
There was a problem hiding this comment.
You put 300 here, whereas it should be 500 if you want to have the same output as for MinCovDet, see comment
|
Thanks @adrinjalali ! |
|
@adrinjalali I think you might have missed my comment about the EllipticEnveloppe issue: you put n_samples=300, whereas it should be 500 if you want to have the same output as for MinCovDet. |
|
@albertcthomas I haven't missed it. The issue I report there is not the inconsistency between the two methods, and is rather just for the EllipticEnvelope. However, if the two methods are supposed to give the same results, it would mean EllipticEnvelope should also be stable if the number of samples is increased to 500. For that I need to do more tests on different systems/servers and haven't had the chance to do so yet. I'll respond to your comment on the issue #12127 once I run the tests. |
See #3846
Adds examples to all
sklearn/covarianceclasses except the ones being handled in PR #11732