3.1.2 Internal-validation and External-validation
The robustness and predictability of the model were evaluated using the internal and external procedure.50The validation parameters such as R 2,Q 2, AARD and RMSE are listed in Table 1. As shown in Figure 6 (a), the good results of LOO-CV with highQ 2 of 0.9965 suggested the model’s stability. For the external validation, the data set was randomly divided into the training set (7216 data points) and the testing set (1804 data points). The calculated and experimental ρ values of the training set and the testing set were plotted in Figure 6 (b). TheR 2 of the training set and testing set were 0.9954 and 0.9964, respectively. The AARD value were 0.56 % and 0.54 %, respectively. Results showed that the ρ (T, P ) model was stable and has good predictability.