Table 2. The four-fold cross-validation results identify GLM as the least powerful and RF as the most powerful.

All models produce an AUC, an optimal threshold (Opt Thresh.), and are capable of identifying the variable that contributes most to prediction (CV), but the machine learning methods often lack standard regression measures.

Model	K-fold	AUC	Opt Thresh.	CV
GLM	1	0.78	0.149	GDD
GLM	2	0.779	0.168	GDD
GLM	3	0.778	0.134	GDD
GLM	4	0.768	0.144	GDD
GAM	1	0.878	0.175	Mean Temp
GAM	2	0.867	0.146	Mean Temp
GAM	3	0.861	0.099	Mean Temp
GAM	4	0.853	0.177	Mean Temp
MaxEnt	1	0.891	0.315	Mean Temp
MaxEnt	2	0.874	0.402	Mean Temp
MaxEnt	3	0.884	0.289	Mean Temp
MaxEnt	4	0.88	0.336	Mean Temp
RF	1	0.938	0.157	Mean Temp
RF	2	0.924	0.197	Mean Temp
RF	3	0.928	0.15	Mean Temp
RF	4	0.919	0.169	Mean Temp