-
Notifications
You must be signed in to change notification settings - Fork 329
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Question - Interpreting Feature contribution #323
Comments
Hi, Feature importance is computed from the sum of local contributions. This means that the feature contains 27% of local contributions. In this case, the feature explains 27% of the price prediction (according to local explainability). The use of this information depends on your use case. For a detailed overview of shapash, you can read this article: |
@ThomasBouche - can you please help me know how cam I find the important features for label 1 and label 0 at the global level? . I dont wish to have a local level feature fpr label 1 and label 0. I wish to know what are the characteristic for label 1 and label 0 respectively |
I went through the SHAPASH documentation https://shapash.readthedocs.io/en/latest/tutorials/tutorial02-Shapash-overview-in-Jupyter.html and came across the below plot
While I understand that the taller/lengthier bar indicates better predictive power. But how do we know whether 0.27 or 0.28 is a good enough feature to consider it as predictive? Is there any range that you can provide which can indicate strong, moderate and weak predictive power features
Does the 1st bar in below graph mean - "overall finish and material of the house contribute 27% to the output variable"? Or is it okay to say that "overall finish and material of the house explain 27% to the variance in the output variable"
The text was updated successfully, but these errors were encountered: