search

ing-bank / skorecard

scikit-learn compatible tools for building credit risk acceptance models
https://ing-bank.github.io/skorecard/
MIT License
85 stars 24 forks source link

add feature selection method #71

Closed JYLFamily closed 1 year ago

JYLFamily commented 3 years ago

Hi skorecad team: 1.How to deal with logistic regression regression feature coefficient is positive(features are transformed by woe)? 2.What is a good way to select scorecard feature to ensure that all feature coefficients are positive? thanks

anilkumarpanda commented 1 year ago

Hi @JYLFamily thanks for the questions .

  1. For the first question ,can you please elaborate on your question ?
  2. For the second one, I don't think there is an automated way to select features such that coefficients are positive . However you can have a robust feature selection process that ensure that you do not have features with high correlation both univariate and multivariate. Despite that sometimes you can end up in a scenario that a feature when used individually produces a positive co-efficient, however when used in a multi-variate setting, leads to a negative co-efficient. That can lead to interpretability issues. One option is to drop those feature/s manually. Generally those are less important features.
anilkumarpanda commented 1 year ago

@JYLFamily we believe this covered by recent commit, hence this issue can be closed. I you have further question please reopen the issue.