Investigation of Factors and Construction of Statistical Models on Predicting Life Casualties in Building Fires in New Taipei City

Abstract

The main purpose of this study is to investigate on finding factors significantly affecting life casualties in building fires and constructing explanatory models for prediction. Factors were firstly generalized from literature review, and then an empirical study using data from fire cases during 2015 to June 2017 in New Taipei City was conducted. Chi-square test and t-test were performed later on to screen out factors closely related to life casualties, and those factors were then put into binary logistic regression analysis stepwise subsequently. The statistically significant variables in the three convergent models (human factors group, ignited and building environmental factors group, and built-in fire safety measures, rescue and fire growth factors group) were physical and mental illness, state of consciousness, horizontal fire compartment, narrow alleys, building structure, fire phases, and hydrant separately. By using the proposed binary logistic models, the overall correct classification rates were up to 81.5, 81.6, and 70.4%, respectively. Ultimately, the study provides suggestion in law enforcement, public education, and fire prevention planning.