Page 42 - ASBIRES-2017_Preceedings

th
Proceedings of the 9 Symposium on Applied Science, Business & Industrial Research – 2017
ISSN 2279-1558, ISBN 978-955-7442-09-9

A Time Series Model to Forecast Dengue Fever Incidences in Kurunegala
District

Kaushalya WAW, Francisco GS
Department of Mathematical Sciences, Wayamba University of Sri Lanka
kaushalyawaw@gmail.com
ABSTRACT

Effect of dangerous Dengue fever has come to all provinces of Sri Lanka. Western
province is the most affected province by Dengue fever among all the provinces. North
Western province has also faced this problem. Between the two districts, Kurunegala
district has the highest effect. The objective of this study was to fit a time series model to
forecast the Dengue fever incidences in Kurunegala district. Methods: Monthly Dengue
patient counts from January 2010 to August 2016 were used for this study. SARIMA (0,1,0)
(0,1,1)6 was fitted as the most suitable model with least Akaike Information Criterion(AIC)
of 859.26 and Mean Absolute Percentage Error (MAPE) of 39.25% using time series
analysis.

KEYWORDS: Dengue, Forecast, Kurunegala, North western, Puttalam

Dengue patient count from January 2010 to
1 INTRODUCTION
August 2016 were collected and used for
North Western Province is one of the this study. R statistical software was used
nine provinces in Sri Lanka. It consists two for the statistical analysis.
districts named as Kurunegala and Puttalam. 2 LITERATURE REVIEW
Between the two districts, Kurunegala
district has reported the highest amount of Gnanapragasam (2016) carried out a
Dengue patients. The data of the most recent time series model to forecast Dengue cases
six years show that the number of Dengue in Colombo Municipal council. SARIMA
patients are more than 1000 in each year in (1, 0, 0) (1, 1, 2)6 was identified as the best
Kurunegala district. Each year there were model with highest R squared value and
2079 average number of cases comes from lowest Akaike Information Criterion (AIC)
Kurunegala district. Good forecast directs to value. Using the final model, the final model
implement correct actions to reduce the risk they presented was forecasted the Dengue
and aware of the future situation. cases from July 2016 to December 2016.
1.1 Objective of the study Gnanapragasam & Cooray (2015) found
statistical models to forecast the Dengue
This study was carried out to identify
the pattern of spread of Dengue fever in cases of Western Province. Two ARMA
North Western Province mainly in models were found for Colombo district and
Kurunegala District, to create a model to Western province. ARMA (2, 5) model was
forecast the Dengue fever incidences in fitted for the Western Province. ARMA (4,
Kurunegala district and forecast the 3) was fitted for the Colombo district.
potential Dengue Patient count for the next Kavinga, Jayakody and Jayasundara
six months. (2013) presented a time series model to
forecast the Dengue patient count in the
1.2 Data Collection
Colombo District. SARIMA (1,1,2) (1,0,1)3
The data were collected from the was found as the best model. Dengue cases
Department of Health Services, North from March 2012 to September 2012 were
Western Province in Sri Lanka. Monthly forecasted. Thus they forecasted the Dengue

32