MIT Press

Skip to Main Content

Table 3

Performance of the best performing adaptive estimators on the four datasets (Muc7_T provides annotation times from two different annotators A and B) trained on the respective train and evaluated on their test splits. We report the mean absolute error (MAE), the rooted mean squared error (RMSE), the coefficient of determination (R²), and Spearman’s ρ.

Name	Model	MAE	RMSE	R²	ρ	t
Muc7_T A	RR	1.87	2.68	0.56	0.80	0.15
	GP	1.79	2.66	0.57	0.82	7.23
	GBM	1.95	2.97	0.47	0.75	3.40
Muc7_T B	RR	2.19	3.42	0.44	0.79	0.02
	GP	2.08	3.37	0.46	0.81	8.85
	LGBM	2.13	3.50	0.41	0.75	2.90
SigIE	RR	7.96	9.50	0.46	0.73	0.00
	GP	7.62	9.60	0.44	0.70	0.08
	GBM	8.22	10.84	0.29	0.55	0.14
SPEC	RR	9.63	13.86	−0.14	0.50	0.03
	GP	7.63	12.07	0.14	0.51	0.73
	GBM	8.05	12.50	0.07	0.35	1.70

Name	Model	MAE	RMSE	R²	ρ	t
Muc7_T A	RR	1.87	2.68	0.56	0.80	0.15
	GP	1.79	2.66	0.57	0.82	7.23
	GBM	1.95	2.97	0.47	0.75	3.40
Muc7_T B	RR	2.19	3.42	0.44	0.79	0.02
	GP	2.08	3.37	0.46	0.81	8.85
	LGBM	2.13	3.50	0.41	0.75	2.90
SigIE	RR	7.96	9.50	0.46	0.73	0.00
	GP	7.62	9.60	0.44	0.70	0.08
	GBM	8.22	10.84	0.29	0.55	0.14
SPEC	RR	9.63	13.86	−0.14	0.50	0.03
	GP	7.63	12.07	0.14	0.51	0.73
	GBM	8.05	12.50	0.07	0.35	1.70