Table 3.

Performance of classification algorithms for 5 categories of social support.

Social support	Results	Naïve Bayesian	Logistic regression	Support Vector Machine (polynomial kernel)	Random forest	Decision tree	AdaBoost
COMa	Accuracy	.696	.787	.783	.771	.767	.804f
	AUC	.839	.817	.768	.848	.75	.852f
PESb	Accuracy	.713	.830	.840f	.830	.81	.817
	AUC	.823	.787	.681	.825f	.687	.817
PISc	Accuracy	.753	.813	.823f	.767	.779	.801
	AUC	.824	.83	.783	.837	.717	.859f
SESd	Accuracy	.893	.901	.970f	.967	.963	.963
	AUC	.749	.867f	.656	.851	.671	.668
SISe	Accuracy	.851	.880	.943f	.931	.937	.914
	AUC	.893f	.803	.745	.86	.766	.869

^aCOM: companionship.

^bPES: providing emotional support.

^cPIS: providing informational support.

^dSES: seeking emotional support.

^eSIS: seeking informational support.

^fdenotes the best performer for each row.