Table 1.

Model evaluation indicators and calculation methods.

Evaluation indicators	Indicator meaning	Computational formula
${\text{R}}_{\text{Is}}^2$	In-sample goodness of fit, in the training set, the model predicts values to the actual observed values	$R_{\mathit{Is}}^2/R_{\mathit{oos}}^2=1-\frac{{\sum }_{i=1}^n{(y_i-\widehat{y_i})}^2}{{\sum }_{i=1}^n{\left(y_i-\overline{y}\right)}^2}$
${\text{R}}_{\text{oos}}^2$	Out-of-sample goodness of fit, in the training set, the model predicts values to the actual observed values
${\mathit{EVS}}_{\mathit{oos}}$	Explanatory variance, in the prediction set, the fit of the degree of variation to the actual observed value	$EV{S}_{\mathit{oos}}=1-\left(var(y-\widehat{y})\right)/\left(var\left(y\right)\right)$
${\mathit{MSE}}_{\mathit{oos}}$	Mean squared error, the expected value of the square between the out-of-sample predicted value and the actual value	${\mathit{MSE}}_{\mathit{oos}}=1/n{\sum }_{i=1}^n{\left(y_i,-,\widehat{y_i}\right)}^2$
${\mathit{MAE}}_{\mathit{oos}}$	Average absolute error, the expected value of the difference between the out-of-sample predicted and actual value	${\mathit{MAE}}_{\mathit{oos}}=1/n{\sum }_{i=1}^n{\left|y_i,-,\widehat{y_i}\right|}^2$
${\mathit{MedAE}}_{\mathit{oos}}$	Absolute median difference, median of the difference between out-of-sample predicted and actual values	${\mathit{MedAE}}_{\mathit{oos}}=\text{median of}\left|y_i,-,\widehat{y_i}\right|$