Socioeconomic drivers of fire risk: County-level risk disparities across the US (2010–2017)

Urban fire has been one of the most significant threats to humans and is not completely avoidable in the modern era. To manage fire risk more effectively, it is critical to identify key factors that influence fire risk. However, conducting proper risk assessments is very challenging due to the complex interactions among these factors and their regional specificity. The current study classifies multiple counties in the United States into normal and rich groups, analyzing the disparities in fire incidents, fire losses, and fire losses/GDP after logarithmically transforming them into two groups: 3.3%, 5.63%, and 8.66%, respectively. Then an ensemble learning model based on MI-RF and MI-SVM-RFE is constructed to identify the key socio-economic factors influencing fire risk in each group. The results show that the MI-RF model demonstrates superior feature selection performance, with an average dimensionality reduction rate of 72.78% and an accuracy rate of 82.9%. The key social-economic factors influencing fire incidents and fire losses in rich cities are 21 and 22, while in normal cities are 14 and 29. The relevant factors affecting the number of fire incidents and the amount of fire losses in each group are not entirely congruent. These findings provide valuable insights for further analysis and interpretation of the underlying reasons for the disparities in fire risk factors, which are related to urban socio-demographic characteristics and economic activities, thereby supporting policymakers in formulating appropriate safety management measures to mitigate fire risk.