[MRG] Fixing test error coming from sklearn 1.8 release#358
[MRG] Fixing test error coming from sklearn 1.8 release#358tgnassou merged 14 commits intoscikit-adaptation:mainfrom
Conversation
skada/metrics.py
Outdated
| kwargs=None, | ||
| ): | ||
| super().__init__() | ||
| super().__init__(None, 1 if greater_is_better else -1, kwargs) |
There was a problem hiding this comment.
For readibility, I would pass the arguments, by name to know to which parameter we are affecting those values.
tgnassou
changed the title
|
Using |
tgnassou
changed the title
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #358 +/- ##
==========================================
+ Coverage 95.91% 95.93% +0.02%
==========================================
Files 66 66
Lines 7223 7236 +13
==========================================
+ Hits 6928 6942 +14
+ Misses 295 294 -1 🚀 New features to boost your workflow:
|
tgnassou
changed the title
rflamary
left a comment
rflamary
left a comment
There was a problem hiding this comment.
That's a very nice update/cleanup. I just have a few small comments
| ): | ||
| if base_estimator is None: | ||
| base_estimator = LogisticRegression(multi_class="multinomial") | ||
| base_estimator = LogisticRegression() |
There was a problem hiding this comment.
this is not needed anymore you are sure? This was used because the output for the prediction of teh scores is needed, please check that it does not break teh computatio,n of teh cose matrix (that it retsiurnes teh same a with mulyinomial)
There was a problem hiding this comment.
With this code the multinomial loss will be used when n>=3. But the code is not breaking since LogisticRegression has a function predict_log_proba.
There was a problem hiding this comment.
wdyt @rflamary ? right now the computation doesn't break.
skada/_subspace.py
Outdated
| torch = self.torch | ||
| if self.torch: | ||
| import torch | ||
| # torch = self.torch |
There was a problem hiding this comment.
do ot keep this as comment
skada/metrics.py
Outdated
| from .utils import check_X_y_domain, extract_source_indices, source_target_split | ||
|
|
||
|
|
||
| # xxx(okachaiev): maybe it would be easier to reuse _BaseScorer? |
There was a problem hiding this comment.
not sure easier but we need to do it now ;) @tgnassou you can cleanup teh comments maybe?
3 participants
No description provided.