MNT move ROSE into RandomOverSampler with addititional parameters

[glemaitre]

closes #792

This PR intends to:

• Move the implementation of ROSE in the RandomOverSampler by adding 2 parameters: smoothed_bootstrap and shrinkage.
• Remove the ROSE class
• Update the documentation by adding links to the literature in RandomOverSampler.
• Update the User Guide mentioning about this parameter.
• Add an example to illustrate the usage of the parameter as well as the limitation -> we cannot resample string anymore.
• Missing a test to check the keys if shrinkage is a dict

[codecov]

Codecov Report

Merging #791 (9b666a0) into master (9b666a0) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #791   +/-   ##
=======================================
  Coverage   98.58%   98.58%           
=======================================
  Files          89       89           
  Lines        5721     5721           
  Branches      477      477           
=======================================
  Hits         5640     5640           
  Misses         80       80           
  Partials        1        1           

---

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 9b666a0...112add4. Read the comment docs.

[pep8speaks]

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

• In the file examples/over-sampling/plot_shrinkage_effect.py:

Line 19:1: E402 module level import not at top of file
Line 20:1: E402 module level import not at top of file
Line 33:1: E402 module level import not at top of file
Line 46:1: E402 module level import not at top of file

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

Line 832:89: E501 line too long (91 > 88 characters)

Comment last updated at 2021-02-12 10:24:27 UTC

[lgtm-com]

This pull request introduces 4 alerts when merging b1d8714 into 9b666a0 - view on LGTM.com

new alerts:

• 4 for Mismatch in multiple assignment

[chkoar]

Is this bounded?

[glemaitre]

nop

[glemaitre]

oh yes negative don't make sense sorry

[glemaitre]

I will update in the new PR.

[lgtm-com]

This pull request introduces 4 alerts when merging 112add4 into 9b666a0 - view on LGTM.com

new alerts:

• 4 for Mismatch in multiple assignment