Table 2.

Results of the cross-validation and the hyperparameters for the final fine-tuned models predicting obesity in the ABCD sub-sample of adolescents with Fitbit and BMI data (N = 2971).

	Evaluation	LassoReg (Lr = 0.08)	Gradient boosted Tree (n_tree = 60, lr = 0.1, depth = 2)	Explainable boosting machine (max_round = 10, Lr = 0.005, max_bin = 16, interaction = 5)
2 × 5 fold cross validation mean performance	AUC	0.735	0.708	0.726
Best performing model	AUC	0.684	0.664	0.737
	ACC	0.86	0.65	0.86