Feature/add mlsmote

[SimonErm]

Reference Issue

The motivation for this PR is mentioned in #340

What does this implement/fix? Explain your changes.

The PR implements MLSMOTE like discribed in Charte, F. & Rivera Rivas, Antonio & Del Jesus, María José & Herrera, Francisco. (2015). MLSMOTE: Approaching imbalanced multilabel learning through synthetic instance generation. Knowledge-Based Systems. -. 10.1016/j.knosys.2015.07.019.

Any other comments?

This implementation is missing lots of validation, sparse matrix support, pandas support and has a bad perfromance. It's alread open because of @chkoar s suggestion in the referenced Issue(#340 ).
Since i am not an experienced python developer i am thankful for every suggestion for improvement

[pep8speaks]

Hello @SimonErm! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file imblearn/over_sampling/_mlsmote.py:

Line 6:1: E302 expected 2 blank lines, found 0
Line 33:80: E501 line too long (100 > 79 characters)
Line 34:80: E501 line too long (102 > 79 characters)
Line 35:70: W291 trailing whitespace
Line 39:80: E501 line too long (89 > 79 characters)
Line 70:1: W293 blank line contains whitespace
Line 87:80: E501 line too long (80 > 79 characters)
Line 102:80: E501 line too long (107 > 79 characters)
Line 125:80: E501 line too long (96 > 79 characters)
Line 126:80: E501 line too long (95 > 79 characters)
Line 156:80: E501 line too long (87 > 79 characters)
Line 163:67: W291 trailing whitespace
Line 182:80: E501 line too long (119 > 79 characters)
Line 184:80: E501 line too long (113 > 79 characters)
Line 196:80: E501 line too long (126 > 79 characters)
Line 240:80: E501 line too long (80 > 79 characters)
Line 247:39: E741 ambiguous variable name 'l'
Line 250:55: E741 ambiguous variable name 'l'
Line 261:15: E741 ambiguous variable name 'l'
Line 279:80: E501 line too long (80 > 79 characters)

Comment last updated at 2020-06-16 17:16:28 UTC

[lgtm-com]

This pull request introduces 5 alerts when merging 948da4a into b861b3a - view on LGTM.com

new alerts:

• 2 for Mismatch in multiple assignment
• 2 for Unused import
• 1 for Unused local variable

[lgtm-com]

This pull request introduces 4 alerts when merging bef0487 into b861b3a - view on LGTM.com

new alerts:

• 2 for Mismatch in multiple assignment
• 2 for Unused import

[codecov]

Codecov Report

Merging #707 into master will decrease coverage by 2.09%.
The diff coverage is 98.65%.

@@            Coverage Diff             @@
##           master     #707      +/-   ##
==========================================
- Coverage   98.65%   96.55%   -2.10%     
==========================================
  Files          82       82              
  Lines        4907     5140     +233     
==========================================
+ Hits         4841     4963     +122     
- Misses         66      177     +111     

Impacted Files	Coverage Δ
imblearn/ensemble/tests/test_forest.py	100.00% <ø> (ø)
imblearn/utils/_show_versions.py	100.00% <ø> (ø)
imblearn/ensemble/_forest.py	97.36% <92.85%> (-0.55%)	⬇️
imblearn/ensemble/_bagging.py	97.82% <94.44%> (-2.18%)	⬇️
imblearn/utils/estimator_checks.py	95.60% <96.34%> (-1.08%)	⬇️
imblearn/_version.py	100.00% <100.00%> (ø)
imblearn/combine/_smote_enn.py	100.00% <100.00%> (ø)
imblearn/combine/_smote_tomek.py	100.00% <100.00%> (ø)
imblearn/datasets/_imbalance.py	88.23% <100.00%> (+1.56%)	⬆️
imblearn/datasets/_zenodo.py	96.77% <100.00%> (+0.10%)	⬆️
... and 56 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 b861b3a...3361578. Read the comment docs.

[lgtm-com]

This pull request introduces 5 alerts when merging 3361578 into 2a0376e - view on LGTM.com

new alerts:

• 3 for Unused local variable
• 2 for Unused import

[aaronbriel]

@SimonErm is this PR still in progress?

[SimonErm]

The current state of the implementation is working for me, but i think it's far from being ready to be merged into this package.
I currently don't have enough time to do a correct integration and i didn't got feedback so far.
I would declare this PR as inactive.

[aaronbriel]

@SimonErm Thanks for the reply.

[rjurney]

I really want this.

[balvisio]

Hi all, I was wondering if someone is working on this or similar implementation of MLSMOTE. I am interested in trying this algorithm. I might have some time to try to implement it. Would anyone be able to review it?

[chkoar]

Hi all, I was wondering if someone is working on this or similar implementation of MLSMOTE. I am interested in trying this algorithm. I might have some time to try to implement it. Would anyone be able to review it?

Contributions are always more than welcome

[balvisio]

@chkoar : Here is a PR that implements MLSMOTE: #927