Table 4.

Predictive performance of random forest models and the confusion matrix of the models.

Models	Predictions	Actual		Sensitivity	Specificity	AUCa
		Positive	Negative
Fecal	Positive	93	14	0.8692	0.7705	0.902
	Negative	14	47
Soil	Positive	20	17	0.7407	0.8784	0.894
	Negative	7	124
WCR-P	Positive	7	5	0.7778	0.8529	0.864
	Negative	2	29

^aAUC is the area under receiver operating characteristic (ROC) curve.