Figure 2: Ensemble regressions improve in quality, up to a point, as the number of bootstraps is increased. | Nature Methods

Figure 2: Ensemble regressions improve in quality, up to a point, as the number of bootstraps is increased.

From: Ensemble methods: bagging and random forests

Figure 2

(a) The consensus regression (blue line) for ten bootstrap iterations (gray lines) for data in Figure 1. (b) Ensemble regressions for 10, 25 and 50 bootstrap iterations. (c) The bagged and OOB errors (εB, εOOB) as a function of the number of bootstraps. The curve is fit to ten simulations at each bootstrap level using locally weighted smoothing. The gray band is the fit's 95% confidence interval.

Back to article page