Table 4.

Predictive models’ performance.

Evaluation matrix			Predictive models
			Logistic regression		RF		DT		SVM
Confusion matrix			Predicted		Predicted		Predicted		Predicted
			Poor	Well	Poor	Well	Poor	Well	Poor	Well
Observed	Poor		3442	2114	4283	1273	4184	1372	3454	2102
	Well		2215	3472	493	5194	536	5151	1425	4262
			%	%			%	%
Accuracy			74.95	96.38			73.73	67.21
Recall			70.49	95.90			66.44	62.49
Sensitivity			70.49	95.90			66.44	62.49
Specificity			80.09	96.80			86.17	73.45
Positive predictive value			80.34	96.41			89.13	75.66
Negative predictive value			70.17	96.35			60.06	59.71
Precision			80.34	96.41			89.13	75.66
F1 score			75.09	96.16			76.13	68.44
Prevalence			53.57	47.25			63.06	56.91
Detection rate			37.76	45.32			41.90	35.56
Detection prevalence			47.00	47.00			47.00	47.00
Balanced accuracy			75.29	96.35			76.30	67.97