Table 2.

Model Performance Metrics (R²—Coefficient of Determination and RMSE—Root mean Square Error) for Random Forest (RF), Artificial Neural Networks (ANNs), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGB) Algorithms on the Models’ Training and Testing of Soil Organic Carbon (SOC), Refractory SOC (RSOC), and ¹³C Isotopic Composition of SOC (δ¹³C_SOC)

Soil C variable	Model	Training		Testing
		RMSE	R2	RMSE	R2
SOC	RF	0.77	0.87	1.47	0.49
	ANN	1.15	0.70	1.49	0.43
	SVM	1.11	0.72	1.66	0.36
	XGB	0.92	0.81	1.53	0.42
δ13CSOC	RF	0.35	0.89	0.79	0.42
	ANN	0.60	0.66	0.85	0.29
	SVM	0.42	0.84	0.88	0.26
	XGB	0.30	0.84	0.81	0.37
RSOC	RF	0.10	0.87	0.19	0.46
	ANN	0.12	0.80	0.18	0.50
	SVM	0.14	0.75	0.20	0.46
	XGB	0.05	0.95	0.20	0.41