You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am also interested in the SAGE value, which measures the contribution to global prediction accuracy, compared to SHAP, which measures the importance of local predictions.
We believe that if SRI decomposition can contribute to prediction accuracy, it will be an important advance in terms of model interpretability and feature selection.
On the other hand, I understand that it is not easy to support SAGE, which is a different concept, but is it likely to be realized?
I am also interested in the SAGE value, which measures the contribution to global prediction accuracy, compared to SHAP, which measures the importance of local predictions.
We believe that if SRI decomposition can contribute to prediction accuracy, it will be an important advance in terms of model interpretability and feature selection.
On the other hand, I understand that it is not easy to support SAGE, which is a different concept, but is it likely to be realized?
https://pypi.org/project/sage-importance/
https://iancovert.com/blog/understanding-shap-sage/
The text was updated successfully, but these errors were encountered: