TABLE 5.

Metrics of the best ML model achieved for each MLQ subscale, both for the validation and the test set.

Subscale	Model	Features (n)			Validation set					Test set
		Eye-tracking	Behavioural	Total	Accuracy	Kappa	AUC	TPR	TNR	Accuracy	Kappa	AUC	TPR	TNR
MLQ-Leadership transformational	kNN	20	3	23	0.78	0.53	0.74	0.8	0.76	0.69	0.4	0.67	0.57	0.83
MLQ-Leadership transactional	Naïve Bayes	13	7	20	0.84	0.66	0.88	0.8	0.9	0.75	0.53	0.83	0.57	1
MLQ-Leadership passive-avoidant	kNN	21	9	30	0.81	0.63	0.86	0.79	0.87	0.67	0.31	0.74	0.71	0.6
MLQ-Leadership laissez	RandomForest	14	0	14	0.74	0.4	0.82	0.88	0.53	0.75	0.31	0.59	1	0.25
MLQ-Leadership effort	kNN	33	5	38	0.84	0.65	0.84	0.8	0.89	0.75	0.47	0.8	0.75	0.75
MLQ-Leadership effectiveness	Naïve Bayes	19	4	23	0.81	0.61	0.8	0.81	0.85	0.67	0.4	0.91	0.5	1
MLQ-Leadership satisfaction	kNN	21	7	28	0.87	0.42	0.82	0.97	0.48	0.92	0.63	0.62	1	0.5
MLQ-Subordinate transformational	kNN	15	4	19	0.78	0.55	0.82	0.76	0.84	0.91	0.82	1	1	0.83
MLQ-Subordinate transactional	kNN	21	2	23	0.82	0.62	0.88	0.79	0.89	0.5	0	0.5	0.5	0.5
MLQ-Subordinate passive-avoidant	kNN	12	7	19	0.81	0.61	0.88	0.79	0.86	0.82	0.65	0.93	0.67	1
MLQ-Subordinate laissez	RandomForest	29	14	43	0.72	0.43	0.82	0.79	0.67	0.67	0.27	0.46	0.86	0.4
MLQ-Subordinate effort	Naïve Bayes	10	5	15	0.86	0.7	0.88	0.87	0.86	0.5	0.08	0.66	0.29	0.8
MLQ-Subordinate effectiveness	kNN	16	2	18	0.68	0.34	0.71	0.74	0.64	0.67	0.33	0.72	0.5	0.83
MLQ-Subordinate satisfaction	kNN	26	12	38	0.8	0.57	0.81	0.81	0.76	0.91	0.81	0.97	1	0.8

The number of variables used by each model is divided according to their source (i.e., eye-tracking, or behavioural data). The values shown per metric in the validation set are the mean values of the cross-validation iterations. TPR and TNR stand for true positive rate and true negative rate, respectively.