search

analyticalmindsltd / smote_variants

A collection of 85 minority oversampling techniques (SMOTE) for imbalanced learning with multi-class oversampling and model selection features
http://smote-variants.readthedocs.io
MIT License
623 stars 138 forks source link

SMOTEWB #69

Closed szghlm closed 1 year ago

codecov[bot] commented 1 year ago

Codecov Report

Merging #69 (5465122) into master (4afb66b) will decrease coverage by 0.97%. The diff coverage is 23.57%.

:exclamation: Current head 5465122 differs from pull request most recent head 21c1d39. Consider uploading reports for the commit 21c1d39 to get more accurate results

@@            Coverage Diff             @@
##           master      #69      +/-   ##
==========================================
- Coverage   99.88%   98.91%   -0.97%     
==========================================
  Files         128      129       +1     
  Lines        9574     9697     +123     
==========================================
+ Hits         9563     9592      +29     
- Misses         11      105      +94
Flag Coverage Δ
unittests ?

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files Coverage Δ
smote_variants/oversampling/_smotewb.py 21.66% <21.66%> (ø)
smote_variants/oversampling/__init__.py 100.00% <100.00%> (ø)
smote_variants/oversampling/_mot2ld.py 100.00% <100.00%> (ø)

:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more

gykovacs commented 1 year ago

Great one! Looking forward to merge this new oversampler into the package!

Due to a recent change in the package structure, new oversampling techniques are not exposed implicitly through queries, and they are not subjected to many of the tests unless they are exposed through queries. Could you please add the new oversampling technique explicitly to the queries to let us see if it passes all edge cases? You need to update the file smote_variants/queries/_queries.py and please also add an estimated runtime to smote_variants/queries/_runtimes.py. The runtime can be anything reasonable as of now, I will rerun all my measurements the next time I add something to the package and update the runtimes with the exact values. Alternatively, you can run the notebook smote_variants/examples/005_speed_test.ipynb and add the result to the _runtimes.py file - although it might not be aligned with all the other results, temporarily it should be OK.

szghlm commented 1 year ago

Thank you for the feedback. The suggested changes have been made.