[WIP] Some test refactoring

[massich]

Reference Issue

What does this implement/fix? Explain your changes.

Any other comments?

[pep8speaks]

Hello @massich! Thanks for updating the PR.

Cheers ! There are no PEP8 issues in this Pull Request. 🍻

Comment last updated on August 18, 2017 at 08:19 Hours UTC

[glemaitre]

I am fine with factorizing some tests. I put it WIP. Change to MRG once that you made a pass on the tests which might be factorized.

[massich]

The changed tests are a single test not multiple therefore are written with an assert within a for. Give it a read.

cc: @glemaitre

[glemaitre]

They were a single test before as well. The difference is that you knew which line was failing but I hope that pytest will tell you the values of score_func and expected_values so it is fine.

[codecov]

Codecov Report

Merging #322 into master will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff          @@
##           master   #322   +/-   ##
=====================================
  Coverage      98%    98%           
=====================================
  Files          66     66           
  Lines        3860   3860           
=====================================
  Hits         3783   3783           
  Misses         77     77

---

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 f02c565...bd0981f. Read the comment docs.

[massich]

Just to have it somewhere (do not comment)

def test_ncr_check_threshold_cleaning_parameter():
    error_regex = "'threshold_cleaning' is a value between 0 and 1"
    for th in [-10, 10]:
        with raises(ValueError, match=error_regex):
            NeighbourhoodCleaningRule(threshold_cleaning=th).fit_sample(X, Y)
    for th in [0, 0.2, 1.]:
        with raises(None):
            NeighbourhoodCleaningRule(threshold_cleaning=th).fit_sample(X, Y)

also test_ratio_float_error():

def test_ratio_minority_under_sampling():
    error_regex = "'ratio'='majority' cannot be used with under-sampler."
    with raises(ValueError, match=error_regex):
        check_ratio('minority', np.array([1, 2, 3]), 'under-sampling')

def test_ratio_majority_over_sampling():
    error_regex = "'ratio'='majority' cannot be used with over-sampler."
    with raises(ValueError, match=error_regex):
        check_ratio('majority', np.array([1, 2, 3]), 'over-sampling')

The errors with the following regex smote needs to be a SMOTE are really wierd.

[massich]

This PR was a place to play with pytest. But now pytest has been in imbalanced-learn for while. What do you guys want to do with it?

BTW, still no consensus regarding with raises(None)

[glemaitre]

We should still update the tests such that they are parametrize and use properly pytest.
Regarding raises(None) I don't think that this is an issue for the moment.