是随机森林的一个有效的替代产品
Stepwise Clustered Ensemble (SCE) is an ensemble of tree-structured models, and it has been used as a useful alternative to the well-known random forest (RF).
在预测精度和可解释性上有优势
In particular, SCE has been reported as a promising data-driven approach for streamflow simulation in terms of predictive accuracy and interpretability (Li et al., 2021a). The basic concept of regression tree ensemble is that when the number of ensemble members (i.e., SCA trees) increases, the variation of the unstable predictions is reduced, thereby increasing the predictive accuracy
Each SCA tree grows in accordance with a random subset of predictors sampled without replacement and a bootstrapped version of t