Open
Description
Atm. ExtraTrees are parsed like sklearn random forest models. Idk if this is 100% valid and we do not have test for ExtraTrees against the shap implementation of TreeSHAP. So I would like to extend the tests similar to our xboost/lightgbm/random forest/decision trees.
If there are errors then we need to fix them.
Metadata
Assignees
Projects
Status
📋 Backlog