Expose monotonic trend analysis to summary table. #297
Conversation
guillermo-navas-palencia
left a comment
guillermo-navas-palencia
left a comment
There was a problem hiding this comment.
Thanks for your contribution. I like the idea of showing the monotonic trend in the summary. Regarding the monotonic score using the Kendall tau or Spearman rho correlation, I think we can choose just one and rename it simply as correlation.
…ion and fix indentation.
|
Hi, Guillermo. Done :- ) |
guillermo-navas-palencia
left a comment
guillermo-navas-palencia
left a comment
There was a problem hiding this comment.
It looks good! Only a minor change and pass flake8.
|
pull failed? |
|
Yes, there are two types of problems:
I would only focus on the first one. I will deal with the second type before merging to master. |
|
@chunqishi I think this would be a nice addition, can you resolve the first issue above? If needed perhaps we could help, but the logs of the last run have expired; maybe just push a merge from master to kick off a new check? |
3 participants
Hi, Guillermo.
Thanks a lot for such great package optbinning.
I used it to variable analysis a lot.
One thing is very cool, this lib has great monotonic analysis.
This monotonic trend results are useful.
please check paper: https://ojs.aaai.org/index.php/AAAI/article/view/7055
This monotonic analysis can used in boost trees by setting monotonic constraints
see: https://xgboost.readthedocs.io/en/stable/tutorials/monotonic.html
see: http://scikit-learn.org/stable/auto_examples/ensemble/plot_monotonic_constraints.html
see: https://catboost.ai/en/docs/references/training-parameters/common#monotone_constraints
Thus: I modified your develop code by 2 parts.
Export the monotonic trend result to binning process summary dataframe.
Add ranking correlation analysis for continuous binning process to double check monotonic trend output.
That's all.
Thank you again.