Page 225 - Contributed Paper Session (CPS) - Volume 2

CPS1834 Gumgum D. et al.
Based on the MSE value for P = 3 the hijri calendar is better than the
Gregorian calendar represented by the SARIMA (1,0,1) model (1,0,1) 12, while
for P = 6 and 12 the Christian calendar is better than the hijri calendar
represented by the same model.
Based on the MAPE value for P = 3 the hijri calendar is better than the
Gregorian calendar represented by the SARIMA (1,0,1) model (1,0,1) 12, while
for P = 6 and 12 the Gregorian calendar is better than the hijri calendar
represented by the same model.Based on the AIC Value for all periods the
Gregorian Calendar is better than the Hijri calendar, for P = 3 and P = 12
represented by the SARIMA (1,0,1) (1,0,1) 12 model while for P = 6 represented
by the SARIMA model (1,0,1) (1,0,2) 12.
From the results of tracking in the previous step, the best model for the
two data above is selected. The two best models are the SARIMA (1,0,1) (1,0,1)
12 and SARIMA (1,0,1) (1,0,2) models 12.
Comparisons are made based on the smallest values of the four criteria
above (MSE, MAD, MAPE and AIC). The smaller the value of the criteria, the
better the model is obtained. The chosen horizon is P = 3.6 and 12 according
to the length of the period.

5. Discussion and conclusion
Both of these data (Gregorian and Hijri) have the same pattern and period,
namely, seasonal patterns and periods of 12.SARIMA modeling based on the
Gregorian Calendar is better than SARIMA based on the Hijri Calendar for
forecasting at 12 and 6 future observations. SARIMA modeling based on the
Hijri Calendar is better than SARIMA based on the Gregorian Calendar for
forecasting on the 3 future observations.

References
1. C.Sivapragasam,shie-yui liong and m.f.k.Pasha (2001),” Rainfall and
Runoff forecasting with SSA-SVM approach”,Journal of
Hydroinformatics,141-152.
2. Asce,upsm and Jothiprakash,V (2015)” Extraction of Nonlinear Rainfall
Trends Using Singular Spectrum Analysis”, Journal of Hydrologic
Engineering,20(12):05015007.
3. C.L Wu and K.W.Chau (2011),”Rainfall-ruoff modelling using artificial
neural network coupled with singular spectrum analysis”,Journal of
Hyidrology,399,394-409.
4. D.Baratta,G Cicion,F.Masulli and L.Studer (2003),”Application of an
ensemble technique based on singular spectrum analysis to daily rainfall
forecasting”,Neural Networks,16,375-387.

214 | I S I W S C 2 0 1 9