Figure 1

Cross-validation of the effect of dataset size. The dataset was randomly divided into five groups. Combinations of four groups were assigned for training the SVR-assisted SRK/T power calculation, and the prediction errors were obtained from the rest of them.