search

awsm-research / PyExplainer

PyExplainer: A Local Rule-Based Model-Agnostic Technique (Explainable AI)
MIT License
29 stars 9 forks source link

Result of the 2nd step of AutoSpearman (multicollinearity mitigation) may be ignored #12

Closed escapar closed 3 years ago

escapar commented 3 years ago

Hello.

I am using python implementation of AutoSpearman, and I found it is not completely removing multicollinearity (i.e., vif > 5 is still presented). Later, I found the result of the 2nd step of AutoSpearman may be ignored in this implementation.

I am referring to this file: https://github.com/awsm-research/PyExplainer/blob/master/pyexplainer/pyexplainer_pyexplainer.py

For example, AS_metrics in line 127: print('Finally, according to Part 2 of AutoSpearman,', AS_metrics, 'are selected.')

are not using results in the while loop (line 125) above.

MichaelFu1998-create commented 3 years ago

Thanks! will update soon as it is solved

MichaelFu1998-create commented 3 years ago

@escapar Hi, I've uploaded a new 1.0.1 version to Pypi with patch for AutoSpearman function, please let me know if there is still any issue about the function, thanks again :) !