Jurnal Biometrika dan Kependudukan
ISSN 2302707X
Vol. 2 / No. 1 / Published : 2013-07
Order : 12, and page :88 - 98
Related with : Scholar Yahoo! Bing
Original Article :
Comparison of methods arima (box jenkins) and method of winter forecasting the number of cases of dengue fever
Author :
- Metta Octora*1
- Kuntoro*2
- Departemen Biostatistika dan Kependudukan Fakultas Kesehatan Masyarakat, Universitas Airlangga, Surabaya
- Departemen Biostatistika dan Kependudukan Fakultas Kesehatan Masyarakat, Universitas Airlangga, Surabaya
Abstract :
A good forecasting method is a method that has the smallest error rate in forecasting. Each ARIMA (Box Jenkins)and Winter method have advantages and disadvantages when compared with other methods. For comparing thesemethods, we used Dengue Haemorrhagic Fever (DHF) case because of seasonal feature. The study has been doneto compare ARIMA and Winter method by determining the best mathematical model, and the smallest predictionerror on the number of DHF cases in Surabaya. The data was DHF case at Health Department of Surabaya forthe period from January 2005 until June 2010. Time series data are classifi ed monthly that are known have cyclicperiodic movements. Earlier variants should be tested fi rst by comparing the individual values with the averagevalue for each year. If the data is already seasonal then analyzed with Winters and ARIMA method. Winters methodused 4 models, while ARIMA method obtained 3 models. Furthermore, mathematical models are determined thesmallest forecasting error rate by the smallest value MAPE, MAD and MSE indicator to predict the incidence ofDHF in the next 6 months. The smallest error in sample value of Winters method is model 3 with MAPE 49.14212;MAD 88.5205; and MSE 18322.02, while the smallest error out sample value of Winters method is a model 4 withMAPE 3.8810; MAD 17.4669 and MSD 4535.979982. The smallest error in sample value of ARIMA method ismodel 1 with MAPE 3.9667, MAD 0.1935 and MSD 0.067899. The smallest error out sample value of ARIMAmethod is model 2 with MAPE 1,0286; MAD 0,0620 and MSD 0.0489032 of the these methods are analyzed can beconcluded that the method of ARIMA (1,0,2) (1,0,2) is the best method because it has the MAPE, MAD and MSDis smaller than the method of Winter with parameters alpha = 0.2, gamma = 0.15 and delta 0.002.
Keyword :
ARIMA, winters, MAPE, MAD, MSD,
References :
Box, G.E.P., & Jenkins, G.M,(1976) Time Series Analysis, Forecasting and Control San Fransisco : Holden-Day
Ispriyanti, D,(2004) Pemodelan Statistika dengan Transformasi Box Cox Jurnal Matematika dan Komputer : Vol. 7. No. 3, 8–17.
Depkes RI,(2005) Pencegahan dan Pemberantasan Demam Berdarah Dengue di Indonesia Jakarta : Ditjen PP & PL
Makridakis, S., Syeven C.W & Victor, E.M,(1995) Metode dan Aplikasi Peramalan, Terjemahan Hari Suminto Jakarta : Binarupa Aksara
Kurniawan, D. ,(2008) Regresi Linier http:// ineddeni.wordpress.com. : (sitasi 1 Februari 2010)
Archive Article
| Cover Media | Content |
|---|---|
 Volume : 2 / No. : 1 / Pub. : 2013-07 |
|
