Table 4.

Mean absolute error (MAE) and root-mean-square error (RMSE) results of proposed experiments using eight different machine learning models with SVR, MLP, CNN, and RNN on emsd, imsd, and custom movement statistics (CMS) features

		emsd				CMS		imsd		SOTA
		SVR	MLP	CNN	RNN	SVR	MLP	SVR	MLP
MAE	val	10.12	5.19	5.03	6.45	8.19	6.75	7.56	5.91	–
	eval	8.60	8.83	8.44	8.13	7.86	8.11	7.31	7.85	8.83
RMSE	val	12.85	6.84	6.55	9.18	10.16	9.04	9.39	8.74	–
	eval	11.56	11.48	10.82	11.41	10.38	10.81	9.56	10.49	12.05

The best overall result is highlighted in bold.