y_min y_max resolved #10981
y_min y_max resolved #10981
Conversation
|
you need to add a test |
|
Is there any place I can find the documentation for reference? |
|
see file test_isotonic.py |
|
I am working on all test cases, will update soon |
|
I agree with @jnothman isotonic is not just for calibrating probabilitiues |
|
The update has been made on calibration, not isotonic.py @agramfort |
sklearn/calibration.py
Outdated
| @@ -390,6 +390,8 @@ def predict_proba(self, X): | |||
|
|
|||
| # Deal with cases where the predicted probability minimally exceeds 1.0 | |||
| proba[(1.0 < proba) & (proba <= 1.0 + 1e-5)] = 1.0 | |||
| proba[proba > 1.0] = 1.0 | |||
There was a problem hiding this comment.
Why are you not using np.clip?
There was a problem hiding this comment.
And isn't the previous line now redundant?
There was a problem hiding this comment.
The python test seems to be failing with maximum calls exceeded
|
Please add an entry to the change log at |
| @@ -141,6 +141,15 @@ Classifiers and regressors | |||
| - :class:`dummy.DummyClassifier` and :class:`dummy.DummyRegresssor` now | |||
| only require X to be an object with finite length or shape. | |||
| :issue:`9832` by :user:`Vrishank Bhardwaj <vrishank97>`. | |||
|
|
|||
| - :class:`neighbors.RadiusNeighborsRegressor` and | |||
There was a problem hiding this comment.
The first seems like a rebase issue.
| # Deal with cases where the predicted probability minimally exceeds 1.0 | ||
| proba[(1.0 < proba) & (proba <= 1.0 + 1e-5)] = 1.0 | ||
| proba[proba > 1.0] = 1.0 | ||
| proba[proba < 0.0] = 0.0 |
There was a problem hiding this comment.
I'm not sure why that line was checking for the eps here. Can you check who wrote that and ping them?
sklearn/tests/test_calibration.py
Outdated
| @@ -239,6 +239,20 @@ def test_sigmoid_calibration(): | |||
| np.vstack((exF, exF)), exY) | |||
|
|
|||
|
|
|||
| def test_isotonic_calibration(): | |||
There was a problem hiding this comment.
you should provide a name relating to what it's testing, i.e. infinite probabilities. Are you sure that case actually happens with the test case you provided?
test_isotonic_calibration_prob_not_inf
for test_isotonic_calibration_prob_not_inf
|
I have added the data set for which the predicted probabilities are "inf", until the commit for the data set is approved, the test for the test_isotonic_calibration_prob_not_inf will not pass. |
| parallelized according to ``n_jobs`` regardless of ``algorithm``. | ||
| :issue:`8003` by :user:`Joël Billaud <recamshak>`. | ||
|
|
||
| - Clipped the values of proba between [0,1] in |
There was a problem hiding this comment.
The repetition of [0,1] is confusing
|
@aishgrt1 We do not include data in the repo. We should find a minimal example which should trigger the issue. Could you authenticate a small dataset to reproduce the error. |
|
I would try to create a similar dataset with "inf" error ! |
Reference Issues/PRs
Fix #10903
What does this implement/fix? Explain your changes.
y_min set to 0 y_max set to 1
Any other comments?