USING MULTIPLE LINEAR REGRESSION TO FORECAST THE NUMBER OF ASTHMATICS Darmesah Gabda 1 , Noraini Abdullah 1 , Kamsia Budin 2 & C.K. Lim 1 1 Programme of Mathematics with Economics 2 Environmental Science Programme Universiti Malaysia Sabah Locked bag 2073, 88999 Kota Kinabalu, sabah MALAYSIA darmesah@yahoo.com , noraini@ums.edu.my , dj2403@yahoo.com / http://www.ums.edu.my Abstract: - The objective of this study was to determine the association between the number of asthmatic patients in Kota Kinabalu, Sabah with the air quality and meteorological factors using multiple linear regression. The main eight independent variables with the fourth order interactions were included in the model. There were 80 possible models considered and the best model was obtained using the eight selection criteria (8SC). The result showed that the best model would represent the cause of the rise in the number of asthmatics modeled by M80.23. Key-Words: - multiple regression, eight selection criteria, fourth-order interaction, best model, asthma. 1 Introduction Asthmatic individuals had been identified as a population that is especially sensitive to the effects of ambient air pollutants [1]. In this study, five criteria pollutants were considered for the assessment of their associations with the number of asthmatics, namely carbon monoxide (CO), ozone (O 3 ), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ) and particular matter (PM 10 ). Meteorological factors such as temperature, relative humidity and rainfall also were considered as contributing causes on the increasing number of asthmatics. Many studies had showed that the different results of asthmatic association with air quality factors and meteorological factors. Hence, the impact of each of these air quality and meteorological factors need to be studied. In this study, multiple regression was used to relate the number of asthmatics with the air quality and meteorological factors. The interaction variables were also included in the model, besides the main variables. In some problems of multiple regression, some independent variables may not be related to the dependent variable. Hence, a procedure to select an appropriate subset independent variable is required to relate with the dependent variable. Criteria on the selection of the best model played an important role in choosing the best model since the total number of variables involved were large. In this study with the help of the eight selection criteria and the level of significance, α equals 0.05, the best model was obtained. Several tests were carried out to the best model such as the individual test, global test, Wald test and randomness test. 2 Literature review Air pollutants had an effect on the respiratory and cardiovascular health although it is still at low level below the national guideline. A study in Klang Valley showed that nitrogen dioxide had the greatest impact on the respiratory and cardiovascular morbidity. It also showed that PM 10 and sulphur dioxide were significantly associated with the relative risks for respiratory and cardiovascular mortality[2]. Stronger associations of coughs among children with PM 10 , PM 2.5 , PM 1 and PM 10-2.5 had been reported [3]. Children were more sensitive to the effects of increased levels of PM air pollution than adults. It was also reported that asthmatic presentations had increased with each 10 / 3 g m μ − increase in PM 10 concentration [4]. However, in some studies, there was no significant association between particulate matter (PM) with the total respiratory admissions and number of children without hyperactivity [5, 6]. The effects of interactions between the air pollutants and meteorological factors towards asthma were also studied. Besides considering the single independent variable as an explanatory to the dependent variable, interaction effects between the independent variable was suggested by [7] to be taken into the model. These 2nd WSEAS Int. Conf on COMPUTER ENGINEERING and APPLICATIONS (CEA'08) Acapulco, Mexico, January 25-27, 2008 ISSN: 1790-5117 Page 256 ISBN: 978-960-6766-33-6