Unemployment rate forecasting: LSTM-GRU hybrid approach

Journal for Labour Market Research

Table 2 Performance results of models

Model	Epoch	MAE	RMSE	MAPE
United States
LSTM	500	0.40	0.97	5.69
	1000	0.33	0.94	4.81
	1500	0.37	0.96	5.30
GRU	500	0.29	0.94	4.56
	1000	0.26	0.93	3.93
	1500	0.29	0.93	4.50
HYBRID	500	0.26	0.93	4.01
	1000	0.31	0.94	4.92
	1500	0.26	0.93	3.91
United Kingdom
LSTM	500	0.31	0.47	6.15
	1000	0.35	0.45	7.22
	1500	0.34	0.43	7.28
GRU	500	0.21	0.30	4.92
	1000	0.27	0.40	6.40
	1500	0.30	0.44	7.13
HYBRID	500	0.20	0.28	4.66
	1000	0.25	0.37	5.89
	1500	0.29	0.42	6.91
France
LSTM	500	0.21	0.26	2.29
	1000	0.20	0.25	2.20
	1500	0.25	0.30	2.70
GRU	500	0.10	0.16	1.19
	1000	0.099	0.16	1.11
	1500	0.10	0.16	1.16
HYBRID	500	0.10	0.16	1.18
	1000	0.096	0.16	1.09
	1500	0.10	0.16	1.18
Italy
LSTM	500	0.57	0.66	5.32
	1000	0.48	0.57	4.43
	1500	0.54	0.63	4.98
GRU	500	0.24	0.35	2.43
	1000	0.24	0.34	2.38
	1500	0.23	0.33	2.27
HYBRID	500	0.24	0.34	2.35
	1000	0.23	0.33	2.28
	1500	0.23	0.33	2.29