Table 5.

Summary of existing representative methods

Category	Method	MAE/year	Accuracy (within 1 year)/%	Dataset scale
Data-driven features	[19]	–	92.29	1248
	[21]	0.41	–	200
	[20]	0.35	–	200
	[20]	0.63	–	80
TW3-C	[28]	0.59	91.4	486
TW3-RUS	[28]	0.55	92.2	486
TW3-RUS	PEARLS	0.45	96.8	1174