Fig. 1: Calibrated and uncalibrated r-statistic and RvE plots for random forest. | npj Computational Materials

Fig. 1: Calibrated and uncalibrated r-statistic and RvE plots for random forest.

From: Calibration after bootstrap for accurate uncertainty quantification in regression models

Fig. 1: Calibrated and uncalibrated r-statistic and RvE plots for random forest.

Distributions of r-statistic values (top row) and RMS residual vs. \(\hat \sigma\) plots (bottom row) for random forest, using the synthetic (panels a and d), diffusion (panels b and e), and perovskite (panels c and f) datasets, shown with both uncalibrated and calibrated \(\hat \sigma\) values. Markers which are not filled in represent bins with fewer than 30 points. Statistics for each plot are summarized in Table 1.

Back to article page