Table 3.
Model validation metrics.
Model | % Variance explaineda | RMSEa | RMSEb | R2 b | MAEb | p valuec |
---|---|---|---|---|---|---|
Temperature | 82.59 | 1.223 | 1.344 | 0.794 | 0.934 | <0.001 |
pH | 81.02 | 0.238 | 0.251 | 0.793 | 0.164 | <0.001 |
Salinity | 85.65 | 4.345 | 4.616 | 0.839 | 2.95 | <0.001 |
Models were tested using a 20% withhold 10-fold cross-validation technique and the randomisation technique recommended by Murphy et al.76 (see Methods for details).
RMSE root-mean square error, MAE mean absolute error.
aCalculated from the “out-of-the-bag” predictions74.
bCalculated using 20% withhold 10-fold cross-validation73.
cCalculated using the randomisation technique76.