The use of machine learning for accidental dwelling fire prevention

In this article the use of machine learning for fire prevention support is examined over the period 2010 to 2024 based on a case study in a fire and rescue service in Northwest England. Machine learning was used to develop a multiple linear regression model of accidental dwelling fire risk at the Lower Super Output Area of geography. This was enhanced by using machine learning to develop a k-means cluster analysis model of communities at the finer grained Output Area level. Over the study period the percentage decrease in accidental dwelling fires in the area studied was 44.2 % compared to a decrease of 27.5 % in England as a whole which appeared to indicate that the more precise targeting of fire prevention resulting from statistical models using machine learning had a positive effect on the effectiveness of fire prevention activities. • Machine learning multiple linear regression model of accidental dwelling fire risk • Machine learning k-means cluster analysis model of fire risk communities • More precise targeting of fire prevention via machine learning • Case study in a fire and rescue service in Northwest England