32911
scikit-learn Docs · نسخهٔ 5e9184e53cd3 · BSD-3-Clause
- The terminal-node update step (a.k.a. line search step) of fitting
sklearn.ensemble.GradientBoostingClassifier was optimized. This mainly benefits deeper trees, which are not the primary use case for gradient boosting but can appear in grid searches. For very deep trees, this can speed up training by up to a factor of 20. By Arthur Lacote <cakedev0>