[MRG] Upgrade to support newest scikit-learn version

[ariapoy]

Reference Issues/PRs

No.

What does this implement/fix? Explain your changes.

When upgrading to sklearn==0.24.2, it occurs five FAILs during python setup.py test.

• Check why test FAILs occurs.
• Update the test case to pass it.

Any other comments?

• It's worth thinking about the efficient ways to upgrade when dependencies upgrade.
• Please see the comments below.

[yangarbiter]

Hi @ariapoy,

Thank you so much for submitting this PR.

There is one thing I would like to make sure, the change of test_hs_active_selecting, test_hs_active_selecting, and test_hs_random_selecting is because of Fixed a bug in datasets.load_iris which had two wrong data points, right?

Thanks. Yao-Yuan

[ariapoy]

Hi @yangarbiter,

Yes! Change of sklearn.dataset.load_iris affect test_hs_active_selecting, test_hs_active_selecting, and test_hs_random_selecting in sklearn==0.20.

[codecov-commenter]

Codecov Report

Merging #188 (35e82ae) into master (2432b51) will increase coverage by 0.12%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #188      +/-   ##
==========================================
+ Coverage   89.47%   89.60%   +0.12%     
==========================================
  Files          37       37              
  Lines        1568     1597      +29     
==========================================
+ Hits         1403     1431      +28     
- Misses        165      166       +1     

Impacted Files	Coverage Δ
libact/query_strategies/multiclass/mdsp.py	78.76% <100.00%> (ø)
libact/utils/__init__.py	77.27% <0.00%> (-4.55%)	:arrow_down:
libact/labelers/ideal_labeler.py	100.00% <0.00%> (ø)
libact/query_strategies/random_sampling.py	100.00% <0.00%> (ø)
...query_strategies/multilabel/binary_minimization.py	100.00% <0.00%> (ø)
..._strategies/multilabel/adaptive_active_learning.py	100.00% <0.00%> (ø)
.../multiclass/active_learning_with_cost_embedding.py	100.00% <0.00%> (ø)
...trategies/density_weighted_uncertainty_sampling.py	98.71% <0.00%> (+0.01%)	:arrow_up:
..._strategies/multilabel/maximum_margin_reduction.py	97.82% <0.00%> (+0.04%)	:arrow_up:
...ery_strategies/multiclass/hierarchical_sampling.py	95.90% <0.00%> (+0.04%)	:arrow_up:
... and 10 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 2432b51...35e82ae. Read the comment docs.

[yangarbiter]

https://github.com/ntucllab/libact/pull/188/checks?check_run_id=3258702214 It appears that sklearn > 0.24 is not installable on python 2.7, maybe we should also consider dropping support for python 2.7.

[ariapoy]

https://github.com/ntucllab/libact/pull/188/checks?check_run_id=3258702214 It appears that sklearn > 0.24 is not installable on python 2.7, maybe we should also consider dropping support for python 2.7.

@yangarbiter Agree! Follow Warning Scikit-learn 0.20 was the last version to support Python 2.7. We could drop support for Python 2.7.

Sorry for forgetting to notify you that I just test Ubuntu:20.04 with default Python 3.8. I notice the FAIL in ubuntu-latest, 3.9 during CI checking. I'll check it later.

Thanks!

[yangarbiter]

https://github.com/ntucllab/libact/pull/188/checks?check_run_id=3258702214 It appears that sklearn > 0.24 is not installable on python 2.7, maybe we should also consider dropping support for python 2.7.

@yangarbiter Agree! Follow Warning Scikit-learn 0.20 was the last version to support Python 2.7. We could drop support for Python 2.7.

Sorry for forgetting to notify you that I just test Ubuntu:20.04 with default Python 3.8. Does it work?

I notice the FAIL in ubuntu-latest, 3.9 during CI checking. I'll check it later.

Thanks!

Libact linting / build (ubuntu-latest, 3.9) (pull_request) this one should fail, which is fine since this is coding style check. I'll try to fix them (you could also help if you want) after this PR is merged.