Timeline for Node Importance in scikit learn
Current License: CC BY-SA 4.0
2 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
May 4, 2021 at 11:04 | comment | added | J.D. | So the weight of the parent node in the equation will always be 1. And when we combine the node importances of all the non-terminal nodes in a tree, there won't be a bias towards say the root node. Ok. As for my second question though, that would reduce to asking whether we are guaranteed the existence of a (nontrivial) splitting that reduces impurity. Is that the case? | |
May 4, 2021 at 9:10 | history | answered | 10xAI | CC BY-SA 4.0 |