search

ombhojane / explainableai

Increase interpretability of your models!
https://pypi.org/project/explainableai/
MIT License
25 stars 40 forks source link

[BUG] InvalidParameterError: The 'estimator' parameter of permutation_importance must be an object implementing 'fit'. Got None instead. #100

Open kooooshhhh opened 1 week ago

kooooshhhh commented 1 week ago

Description

Getting this error when implementing the code for "Multi-Model Comparison" Example in the readme file when implementing this on my own dataset.

Steps to Reproduce

Run the code "Multi-Model Comparison" Example in the readme file by inputing your own csv

Expected Behavior

To provide a pdf report output.

Actual Behavior

InvalidParameterError: The 'estimator' parameter of permutation_importance must be an object implementing 'fit'. Got None instead.

The issue is arising on the results = xai.analyze() function. I think the _calculate_feature_importance function should be checked upon.

kooooshhhh commented 1 week ago

This works when implementing this with another dataset (diabetes prediction) the report is generated

ombhojane commented 1 week ago

you want to fix this issue? @kooooshhhh

kooooshhhh commented 1 week ago

Sure, but i think theres a problem with the dataset itself. Thus will need to make the preprocessing pipeline more robust catering to all types of structured datasets. I have a few ideas and code to overcome this issue, would like to discuss this with you

ombhojane commented 1 week ago

yeah sure

kooooshhhh commented 5 days ago

can u provide me with your email so i could reach out to you