Time series model for forecasting the number of new admission inpatients

Table 5 Prediction performance results of three models

Time	Model	Modeling error			Testing error
series	Model	RMSE	MAE	MAPE*	RMSE	MAE	MAPE*
Monthly	SARIMA	759.67	573.75	6.81	426.45	410.97	4.56
	NARNN	448.04	232.92	2.73	886.39	692.29	7.61
	SARIMA-NARNN	433.82	299.22	3.48	378.03	327.75	3.64
Daily	SARIMA	73.73	42.01	18.50	37.29 ^a	20.36 ^a	4.91 ^a
					87.01^b	51.11 ^b	24.22 ^b
	NARNN	32.97	23.49	9.42	24.54 ^a	14.12 ^a	3.44 ^a
					86.88 ^b	49.24 ^b	23.14 ^b
	SARIMA-NARNN	36.86	25.59	10.37	35.82 ^a	18.76 ^a	5.15 ^a
					98.24 ^b	63.43 ^b	33.90 ^b

*MAPE values should be multiplied by 10⁻². ^athe testing error in one-week, ^bthe testing error in four-weeks

ISSN: 1472-6947