Fig. 2

Parity density plots for experimentally determined and predicted values of the validation data (N = 3,076) for each of the six base regression models. The linear models, except for Lasso, appear to have a similar performance while the Random Forest shows a better visual fit and higher density of points along parity line. a Linear, b Ridge, c Lasso, d Random Forest, e Gradient Boosted, f k-Nearest Neighbors