Table 1.

Comparison of candidate SARIMA models.

Model	Estimate	p-Value	Ljung–Box Q Test			AIC	BIC	RMSE	MAPE
			Statistics	DF	p-Value
SARIMA(0,1,1)(0,1,0)[12]			19.42	17	0.305	−240.68	−235.87	0.050	14.008
MA1	−0.868	<0.001
SARIMA(0,1,1)(1,1,2)[12]			14.96	15	0.454	−240.51	−228.48	0.046	12.249
MA1	−0.882	<0.001
SAR1	−0.780	<0.001
SMA1	0.567	0.046
SMA2	−0.433	0.044
SARIMA(1,0,1)(0,1,0)[12]			19.40	16	0.249	−249.89	−242.63	0.048	12.411
AR1	−0.777	<0.001
MA1	1.000	<0.001
SARIMA(1,0,1)(0,1,1)[12]			15.45	15	0.420	−253.67	−244.00	0.045	11.888
AR1	−0.751	<0.001
MA1	1.000	<0.001
SMA1	−0.398	0.019
SARIMA(1,0,1)(1,1,0)[12]			16.85	15	0.328	−252.53	−242.85	0.046	12.178
AR1	−0.746	<0.001
MA1	1.000	<0.001
SAR1	−0.297	0.025
SARIMA(1,0,1)(1,1,1)[12]			13.26	14	0.506	−252.07	−239.98	0.045	11.754
AR1	−0.759	<0.001
MA1	1.000	<0.001
SAR1	0.289	0.477
SMA1	−0.688	0.084
SARIMA(2,0,2)(0,1,0)[12]			18.18	16	0.313	−247.64	−235.55	0.047	12.780
AR1	−1.435	<0.001
AR2	−0.925	<0.001
MA1	1.581	<0.001
MA2	1.000	<0.001
SARIMA(2,0,2)(0,1,1)[12]			16.67	15	0.339	−251.68	−237.17	0.045	12.266
AR1	−1.067	<0.001
AR2	0.950	<0.001
MA1	1.907	<0.001
MA2	0.830	<0.001
SMA1	−0.500	<0.001
SARIMA(3,0,0)(0,1,0)[12]			15.10	17	0.588	−256.66	−246.98	0.045	12.564
AR1	0.062	0.536
AR2	−0.018	0.859
AR3	0.469	<0.001

AIC: Akaike information criterion; BIC: Bayesian information criterion; RMSE: root mean squared error; MAPE: mean absolute percent error; DF: degree of freedom.