Fig. 4: Comparison of the accuracy of the machine learning models.

Predicted vs. measured VHT alteration rate for (a) PQM, (b) ANN, (c) LLR, and (d) GPR models, based on the same test samples from a single train-test split. The points are color-coded based on the density of the scatters on the plot (the concentration increases from blue to red), and * represents the unit of the scaled VHT alteration rate, i.e., g/(m2·d). The dashed line is added to visualize the perfect agreement.