Metabolic Syndrome Prediction Models Using Machine Learning and Sasang Constitution Type

Table 4

Order of informative value of variables associated with metabolic syndrome as identified by the random forest method.

Total		Sasang constitution
		TE (n = 1,342)		SY (n = 795)		SE (n = 734)
Variable	Importance	Variables	Importance	Variables	Importance	Variables	Importance
BMI	100.0	BMI	100.0	BMI	100.0	BMI	100.0
Age	49.4	Age	58.0	Age	64.6	Age	43.5
Stress	37.4	Stress^†	43.8	Stress	53.7	Stress	42.6
Activity	24.2	Activity^†	28.1	Alcohol consumption	35.8	Activity	25.2
Alcohol consumption	23.2	Alcohol consumption	26.2	Activity	35.0	Alcohol consumption	18.9
Sasang constitution^†	16.0	Education	9.7	Education	13.1	Education	10.2
Education	8.7	Smoking^†	4.7	Smoking	7.4	Marital status^†	4.2
Smoking	3.8	Sex	1.6	Sex	2.3	Smoking	2.2
Sex^†	1.1	Marital status	0.0	Marital status	0.0	Sex	0.01
Marital status	0.0	—	—	—	—	—

and ^† in logistic regression. BMI, body mass index; SE, So-Eum; SY, So-Yang; TE, Tae-Eum.