@JeanMoneger It is actually not. Here’s more info: https://t.co/P08Wz9encj. We actually modeled random noise one time in an unpublished paper and it matched quite well: https://t.co/skMS5ZFiGr
3,548 followers
606 followers
RT @introspection: ⚠️ Random Forest variable importances have a bias towards correlated predictors because of tree building process & compu…
6,708 followers
RT @introspection: ⚠️ Random Forest variable importances have a bias towards correlated predictors because of tree building process & compu…
8,862 followers
⚠️ Random Forest variable importances have a bias towards correlated predictors because of tree building process & computation of the importance measure 💡 Solution: conditional permutation scheme: https://t.co/xdfPJhNBbT #DataScience cc @DaniloBzdok