Adaptive-Lasso Analysis For The Number Of Urban Firesand Meteorological Factors

With the increasing stern situation of fire disaster in China, every year there will be up to tens of billions of economic losses caused by fires and hundreds and even thousands of casualties, which exerts a bad social impact among the people. Therefore, it is essential to explore the objective principles of the occurrence and development of fires and set up efficient disaster early-warning mechanism. In this thesis, the fires and the meteorological data in Tianjing City between 1997 and 2007 were taken as the research subject. The research method of Adaptive-Lasso Method was employed to study the relationship between the frequency of the urban fires and the meteorological elements.This thesis consists of four chapters. In Chapter One, an introduction was presented, including the review on the previous studies on fires in recent years and a summary of the articles with the fires and the meteorological data in Tianjing City between 1997 and 2007 as the research subject.Chapter two mainly states the methodology of the research. Adaptive-Lasso method was employed in this research, which is of certain advantage on multivariate modeling and variable selection.In Chapter Three, a linear regression model between the monthly fire frequency and monthly meteorological elements, such as temperature, wind speed, rainfall, sunshine, humidity, etc and the number of fire in last month was set up. With the Adaptive-Lasso method, the above mentioned variables were selected and their corresponding parameters were estimated in this thesis; moreover, the obtained model was analyzed and compared and the predictive analysis of the fire frequency in the first half year in 2008 was made. The results showed that the linear model set up with Adaptive-Lasso method is beneficial to better predict the fire frequency.In Chapter Four, a conclusion of the thesis was presented. Due to the fact that meteorological elements are not the unique reason to fires, it is essential to take other elements into account, such as economic elements and the distribution of combustible materials, etc.