方法对比

	CatBoost	正则化梯度提升	正则化 LightGBM
领域	机器学习	机器学习	机器学习
方法族	Machine learning	Machine learning	Machine learning
起源年份	2018	2001 (gradient boosting); 2016 (explicit L1/L2 regularization in XGBoost)	2017
提出者	Prokhorenkova, L. et al. (Yandex)	Chen, T. & Guestrin, C. (building on Friedman, J. H.)	Ke, G. et al. (Microsoft Research)
类型	Gradient boosting on decision trees	Regularized ensemble (additive tree model)	Regularized gradient boosting ensemble
开创性文献	Prokhorenkova, L., Gusev, G., Vorobev, A., Dorogush, A.V. & Gulin, A. (2018). CatBoost: Unbiased Boosting with Categorical Features. In NeurIPS 2018.	Chen, T. & Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794.	Ke, G., Meng, Q., Finley, T., Wang, T., Chen, W., Ma, W., Ye, Q., & Liu, T.-Y. (2017). LightGBM: A highly efficient gradient boosting decision tree. Advances in Neural Information Processing Systems, 30, 3146–3154.
别名	CatBoost (Categorical Boosting), categorical boosting, ordered boosting, kategorik gradyan artırma	penalized gradient boosting, shrinkage-regularized boosting, XGBoost-style regularization, L1/L2 gradient boosting	LightGBM with L1/L2 regularization, penalized LightGBM, LightGBM ridge/lasso, regularized LGBM
相关	5	6	5
摘要	CatBoost is a gradient boosting algorithm, introduced by Prokhorenkova and colleagues at Yandex in 2018, that handles categorical variables natively and uses ordered target encoding to avoid label leakage. By building an additive ensemble of trees while permuting the data order at each iteration, it is often superior to XGBoost and LightGBM on category-heavy data.	Regularized gradient boosting extends the classic additive tree ensemble (Friedman 2001) by embedding L1 and L2 penalty terms directly into the training objective, along with a complexity penalty on tree size. Popularized by XGBoost (Chen & Guestrin 2016), this framework reduces overfitting and improves generalization compared to unpenalized boosting, while retaining the method's characteristic accuracy on tabular data.	Regularized LightGBM applies L1 (lasso) and L2 (ridge) penalty terms to the leaf weight objective of LightGBM — Microsoft's highly efficient gradient boosting framework — to control model complexity, reduce overfitting, and improve generalization on tabular classification and regression tasks with high-dimensional or noisy feature sets.