Time Series Models to Forecast Dengue Fever Incidences in Western Province of Sri Lanka
Abstract
Dengue is one of the most dangerous mosquito viral infections in the world. In the recent past dengue has become the number one killer mosquito infection in Sri Lanka. Each year the number of incidences of dengue is dramatically increasing. According to the records in Health Ministry of Sri Lanka, 47246 dengue fever incidences are reported in 2014. Since awareness of dengue infections is of utmost importance, Centre for Dengue Research at the Department of Microbiology of Faculty of Medical Sciences of University of Sri Jayewardenepura has been established in 2012 by the government of Sri Lanka. The descriptive statistics of district wise data indicate that more cases are reported from Colombo district in Western province. Therefore the objective of this study is to fit models for Colombo district in particular and Western province of Sri Lanka in general. Accordingly, national level arrangements can be implemented by relevant body to control the incidences as well as to take necessary action in preparation for treatments. The relevant tests in time series analysis are carried out to develop two ARMA models, one for Western Province, and one for Colombo district. The data, from January 2010 to December 2014, published by Health Ministry of Sri Lanka are taken to develop models. The fitted models are used to forecast next-three months incidences from January to March 2015. To check the validation of models Augmented Dickey-Fuller test, Boxpierce Q statistic, serial correlation test and White heterosedasticity test are used. MAPE statistics is used to get accuracy of fitted models. Through this study, it can be confirmed that 50% of total incidences are reported in Western province of Sri Lanka. Further, it is confirmed that one fourth of the total incidences are from Colombo district in Western province. MAPE statistics show nearly 15% and thus the fitted models are acceptable for forecast the incidences. Dengue is one of the most dangerous mosquito viral infections in the world. In the recent past dengue has become the number one killer mosquito infection in Sri Lanka. Each year the number of incidences of dengue is dramatically increasing. According to the records in Health Ministry of Sri Lanka, 47246 dengue fever incidences are reported in 2014. Since awareness of dengue infections is of utmost importance, Centre for Dengue Research at the Department of Microbiology of Faculty of Medical Sciences of University of Sri Jayewardenepura has been established in 2012 by the government of Sri Lanka. The descriptive statistics of district wise data indicate that more cases are reported from Colombo district in Western province. Therefore the objective of this study is to fit models for Colombo district in particular and Western province of Sri Lanka in general. Accordingly, national level arrangements can be implemented by relevant body to control the incidences as well as to take necessary action in preparation for treatments. The relevant tests in time series analysis are carried out to develop two ARMA models, one for Western Province, and one for Colombo district. The data, from January 2010 to December 2014, published by Health Ministry of Sri Lanka are taken to develop models. The fitted models are used to forecast next-three months incidences from January to March 2015. To check the validation of models Augmented Dickey-Fuller test, Boxpierce Q statistic, serial correlation test and White heterosedasticity test are used. MAPE statistics is used to get accuracy of fitted models. Through this study, it can be confirmed that 50% of total incidences are reported in Western province of Sri Lanka. Further, it is confirmed that one fourth of the total incidences are from Colombo district in Western province. MAPE statistics show nearly 15% and thus the fitted models are acceptable for forecast the incidences.