[MRG] ENH: safe-level SMOTE

[laurallu]

This is an implementation of the safe-level SMOTE proposed in the following paper:

C. Bunkhumpornpat, K. Sinapiromsaran, C. Lursinsap, "Safe-level-SMOTE: Safe-level-synthetic minority over-sampling technique for handling the class imbalanced problem," In: Theeramunkong T.,
Kijsirikul B., Cercone N., Ho TB. (eds) Advances in Knowledge Discovery and Data Mining. PAKDD 2009. Lecture Notes in Computer Science, vol 5476. Springer, Berlin, Heidelberg, 475-482, 2009.

Todo list:

• add unit tests

[lgtm-com]

This pull request introduces 1 alert when merging bcc3069 into 321b751 - view on LGTM.com

new alerts:

• 1 for Redundant comparison

[codecov]

Codecov Report

Merging #626 into master will increase coverage by 0.05%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #626      +/-   ##
==========================================
+ Coverage   97.93%   97.98%   +0.05%     
==========================================
  Files          83       84       +1     
  Lines        4784     4911     +127     
==========================================
+ Hits         4685     4812     +127     
  Misses         99       99

Impacted Files	Coverage Δ
imblearn/over_sampling/_smote.py	97.73% <100%> (+0.52%)	⬆️
imblearn/over_sampling/__init__.py	100% <100%> (ø)	⬆️
...blearn/over_sampling/tests/test_safelevel_smote.py	100% <100%> (ø)

---

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 afbf781...866a04f. Read the comment docs.

[glemaitre]

Thanks for the contribution.

You will need to add tests to check that the new function is giving expected results.

[glemaitre]

Oh I see that you mentioned it now :)

[laurallu]

I just added some tests. Any suggestions?

[lgtm-com]

This pull request introduces 1 alert when merging 394d686 into 321b751 - view on LGTM.com

new alerts:

• 1 for Redundant comparison

[chkoar]

@glemaitre SafeLevelSMOTE vs SLSMOTE

[chkoar]

I would use the name _assign_safe_levels unless it hurts the readability in the calling code.

[chkoar]

Thanks!

I just added some tests. Any suggestions?

1. Please use full names in variables, if you can. E.g. sl should be safe_lavels. Unless it hurts the readability.
2. Could you add a section in the documentation?

[lgtm-com]

This pull request introduces 1 alert when merging fd11e32 into afbf781 - view on LGTM.com

new alerts:

• 1 for Redundant comparison

[laurallu]

1. Please use full names in variables, if you can. E.g. [`sl`](https://github.com/scikit-learn-contrib/imbalanced-learn/blob/394d686364725763de8ea2cc3f504d8c08fe111a/imblearn/over_sampling/_smote.py#L1469) should be `safe_lavels`. Unless it hurts the readability.

2. Could you add a section in the [documentation](https://github.com/scikit-learn-contrib/imbalanced-learn/blob/master/doc/over_sampling.rst)?

I've made changes accordingly. I think it's probably ready to go through a detailed review.

[glemaitre]

I would suggest moving this implementation into smote_variants. The idea behind this move is to benchmark the smote variants on a common benchmark on a large number of datasets and include in imbalanced-learn only the versions that show an advantage. You can see the discussion and contribute to it: https://github.com/gykovacs/smote_variants/issues/14

@laurallu would this strategy would be fine with you?

[chkoar]

Since, Safe Level SMOTE exists there IMHO I believe that we should review @laurallu PR and merge it in imblearn.

[glemaitre]

OK, let's do that. Let's open an issue to discuss the inclusion criterion to explain what we are expecting in the future. I will review this PR in a near future.

[laurallu]

Thanks for pointing out the smote_variants to me. I will check it out. I would love to see the inclusion criterion too since I might code up more methods.