marcotcr / lime

Lime: Explaining the predictions of any machine learning classifier
BSD 2-Clause "Simplified" License
11.5k stars 1.79k forks source link

lime for all samples #557

Closed amaa11 closed 3 years ago

amaa11 commented 3 years ago

Dear all,

Assuming I have 200 samples and 11 features. I want to apply lime for all samples and then mean their ranks to have like a global lime for all samples (as the case in SHAP). Is that possible? I tried to output the lime as map and list in lime, but it is hard to extract the ranks and find them mean for all samples. Any idea? Does anyone did it before? Regards

marcotcr commented 3 years ago

I don't understand why it is hard, so I'll close the issue : )