Compatibility with scikit-learn v0.16.1 #243
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Remove SVR coefficient correction (#111) now that the original bug has been fixed in sklearn. - Remove dependency on `BaseLibLinear` in `model_params()` since it is no longer exposed in scikit-learn. - Expose `LinearSVR` from scikit-learn and so do not have SVR have a 'linear' kernel by default. - Include `RescaledLinearSVR` among the rescaled regressors. - Remove `SVR` from `test_linear_models()` in `test_regression.py` and create a new `test_non_linear_models()`.
- Update one of the expected values in `test_sparse_predict_sampler()` likely due to the following item in the scikit-learn 0.16 release notes: "RBFSampler with gamma=g formerly approximated rbf_kernel with gamma=g/2.; the definition of gamma is now consistent, which may substantially change your results if you use a fixed value. " - Fix typo in comment.
- Remove `SVR` from `test_linear_models()` in `test_regression.py` and create a new `test_non_linear_models()`.
- Update expected values in `test_scaling()` likely due to this item in the scikit-learn v0.16 release notes: "Fix numerical stability issues in linear_model.SGDClassifier and linear_model.SGDRegressor by clipping large gradients and ensuring that weight decay rescaling is always positive".
- Since we do not specify 'linear' as the deafult kernel anymore.
|
@dan-blanchard and @mheilman, I assigned @aoifecahill because I can only assign one person to a PR but can you guys also take a look, if you can spare the time? |
- Increasing test coverage.
|
I am not going to spend any more time worrying about the coverage here since (a) I can't figure out what exactly what lines are not covered anymore (coveralls.io shows all lines that are not covered, not just the ones that lost coverage), and (b) because the decrease is so small. |
|
Looks fine to me :) |
Conflicts: .travis.yml requirements_rtd.txt
desilinguist
added a commit
that referenced
this pull request
Jul 12, 2015
…n_compatibility Compatibility with scikit-learn v0.16.1
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
scikit-learnto version 0.16.1 in.travis.yml,requirements.txt, andrequirements_rtd.txt.BaseLibLinearinmodel_params()since it is no longer exposed in scikit-learn (skll breaks importing missing class BaseLibLinear with scikit-learn 0.16 #233, Changes based on big LibLinear refactoring in scikit-learn #235).LinearSVRfrom scikit-learn and so changeSVRto not have a 'linear' kernel by default.RescaledLinearSVRamong the rescaled regressors.SVRfromtest_linear_models()intest_regression.pyand create a newtest_non_linear_models().test_sparse_predict_sampler()likely due to the following item in the scikit-learn 0.16 release notes: "RBFSampler with gamma=g formerly approximated rbf_kernel with gamma=g/2.; the definition of gamma is now consistent, which may substantially change your results if you use a fixed value."test_scaling()likely due to this item in the scikit-learn v0.16 release notes: "Fix numerical stability issues in linear_model.SGDClassifier and linear_model.SGDRegressor by clipping large gradients and ensuring that weight decay rescaling is always positive."