Table 11.

Best 5 models for discriminating the top 3 from the remaining pool of presenting problems.

Number of predictors and predictor names			AUCa		General accuracy (%)		Average accuracyb (%)		F1 score
1
	Clout score	0.70		64.9		86		0.68
	Social processes	0.68		62.2		84.2		0.64
	Authenticity score	0.66		61.4		81.5		0.67
	First-person singular pronouns	0.66		62.7		84		0.65
	Word count	0.64		56.8		N/Ac		0.45
2
	Clout score+discrepancy	0.74		66.9		80.4		0.70
	Clout score+functional	0.73		67.3		82.5		0.70
	Clout score+future focus	0.73		67		81		0.70
	Clout score+drives	0.73		66		82.9		0.68
	Clout score+insight	0.73		68.2		83.6		0.69
3
	Clout score+discrepancy+focus future	0.75		67.1		77.6		0.70
	Clout score+drives+future focus	0.75		67.8		79.4		0.70
	Insight+social processes+functional	0.74		67.8		81.8		0.69
	Clout score+authenticity score+future focus	0.74		67		78		0.69
	Clout score+insight+drives	0.74		68.3		81.5		0.70
4
	Clout score+positive emotions+discrepancy+future focus	0.76		67.9		77.7		0.71
	Clout score+emotional tone score+discrepancy+future focus	0.76		66.9		78.6		0.70
	First-person singular pronouns+discrepancy+social processes+future focus	0.75		67.7		77.4		0.70
	Clout+first-person singular pronouns+discrepancy+future focus	0.75		68.3		77.8		0.71
	Clout+insight+drives+future focus	0.75		68.4		79.1		0.70
5
	Clout score+emotional tone score+discrepancy+functional+future focus	0.76		68.8		76.5		0.70
	Clout score+emotional tone score+discrepancy+drives+future focus	0.76		67		77.3		0.71
	Clout score+emotional tone score+discrepancy+social processes+future focus	0.76		68.4		77.9		0.71
	Clout score+emotional tone score+insight+discrepancy+future focus	0.76		68.3		77.8		0.71
	Word count+clout score+emotional tone score+discrepancy+ffuture focus	0.76		67.7		78.5		0.70
21d	All predictors	0.72		63.8		67.9		0.66

^aAUC: Area under the Receiver Operating Characteristic curve.

^bAverage accuracy when the predicted probability threshold is set to 70%, 80%, and 90%.

^cN/A: not applicable.

^dThe best models with 6-20 predictors (total of 75 items) have lower AUC, general accuracy, average accuracy, and F1 score, thus are omitted.