Table 2. Tuberculosis incidence forecasting until 2025 using the SARIMA and SARIMA-ETS-hybrid models.

Ahead time	Actual-value (cases per 105-month	Models	Prediction (cases per 105-month)	95% CI of PV	RMSE	MAE	MAPE	MASE
3 months 2017-09	3.55	SARIMA(1,0,0)(2,1,0)12	3.43	[2.77–4.14]	0.086	0.058	0.575	0.753
		ETS	3.49	[2.93–4.14]	0.077	0.058	0.569	0.746
		Hybrid	3.39	[2.77–4.13]	0.081	0.056	0.554	0.726
6 months 2017-12	3.3	SARIMA	3.34	[2.76–4.15]	0.069	0.047	0.459	0.606
		ETS	3.29	[2.76–3.97]	0.060	0.046	0.449	0.593
		Hybrid	3.4	[2.79–4.15]	0.064	0.045	0.445	0.587
12 months 2018-6	3.15	SARIMA	3.58	[2.9–4.51]	0.088	0.062	0.598	0.606
		ETS	3.47	[2.9–4.14]	0.079	0.061	0.586	0.785
		Hybrid	3.69	[3.03–4.51]	0.082	0.060	0.578	0.776
18 months 2018-12	3.15	SARIMA	3.16	[2.55–4.0]	0.090	0.069	0.668	0.898
		ETS	3.09	[2.55–3.73]	0.080	0.065	0.627	0.849
		Hybrid	3.23	[2.62–4.0]	0.084	0.066	0.639	0.859
24 months 2019-6	3	SARIMA	3.4	[2.64–4.4]	0.092	0.074	0.714	0.960
		ETS	3.26	[2.64–4.02]	0.079	0.065	0.627	0.812
		Hybrid	3.41	[2.64–4.4]	0.084	0.068	0.655	0.881
30 months 2019-12	3	SARIMA	2.89	[2.38–3.51]	0.091	0.073	0.699	0.942
		ETS	2.90	[2.28–3.69]	0.079	0.064	0.619	0.833
		Hybrid	2.98	[2.28–3.9]	0.084	0.067	0.646	0.870

Notes:

Evaluation of the models’ accuracy of the SARIMA. ETS and SARIMA-ETS hybrid in forecasting underlying trend (pseudo out-of-sample) performance.

Prediction: TB incidence (cases per 100,000-month); Hybrid: SARIMA-ETS-hybrid.

95% CI of PV: 95% confidence interval of predictive value for TB incidence.