Table 1.

Mean absolute relative error of out of series prediction at forecasting horizons of 1 to 4 months ahead for districts in Sri Lanka for the best (S)ARIMA model tested.

District	Horizon 1		Horizon 2		Horizon 3		Horizon 4
	Criterion	Model (pdqPDQ)	Criterion	Model (pdqPDQ)	Criterion	Model (pdqPDQ)	Criterion	Model (pdqPDQ)
Ampara	0.37	012SOH	0.48	012101	0.58	012101	0.60	012101
Anuradhapura	0.23	211101	0.37	210110	0.45	012110	0.51	210110
Badulla	0.43	110SOH	0.62	111SOH	0.75	212101	0.74	112100
Batticaloa	0.36	010011	0.54	012101	0.66	012101	0.78	012101
Colombo	0.35	011000	0.38	112000	0.43	211001	0.46	011000
Galle	0.49	212002	0.58	211101	0.63	211101	0.71	211110
Gampaha	0.40	011111	0.56	011SOH	0.67	011SOH	0.78	011SOH
Hambantota	0.31	010101	0.47	110101	0.60	210101	0.71	210101
Jaffna	0.42	010011	0.58	012111	0.74	012011	0.82	012SOH
Kalutara	0.54	112100	0.72	011000	0.79	110000	0.79	110000
Kandy	0.33	012101	0.43	012101	0.48	112SOH	0.51	212SOH
Kegalle	0.37	010SOH	0.55	211011	0.66	211SOH	0.75	211SOH
Kilinochchi	0.51	010101	0.95	010101	2.13	111010	2.13	010002
Kurunegala	0.25	011011	0.41	010011	0.53	011011	0.63	011011
Mannar	1.16	011100	0.97	012101	1.10	112100	1.18	111101
Matale	0.37	110101	0.53	110101	0.62	212011	0.70	112011
Matara	0.35	212101	0.40	011101	0.46	212101	0.49	0110111
Moneragala	0.29	110100	0.40	011100	0.48	210100	0.56	011100
Mullaitivu	1.03	111100	1.70	112000	2.00	110000	2.58	111SOH
Nuwara Eliya	0.48	212111	0.58	212101	0.66	212101	0.68	111000
Polonnaruwa	0.32	111101	0.47	012101	0.57	111011	0.66	111011
Puttalam	0.35	010101	0.46	010101	0.60	212101	0.72	010101
Ratnapura	0.30	011111	0.43	012111	0.50	210111	0.57	112111
Trincomalee	0.53	112000	0.79	010100	1.05	010100	1.15	112111
Vavuniya	1.22	012000	1.43	012101	1.41	211101	1.48	012101

Legend: pdq = order of autoregressive component, integrated component and moving average component; PDQ = order of seasonal autoregressive component, seasonal integrated component and seasonal moving average component; SOH = seasonality through second order harmonic;