Patterns of fatal road crashes in different road types: applying association rules mining in police reported crash data

Road crashes and resulting fatalities are a major concern globally. Low- and Medium-Income Countries (LMIC) contribute to nearly 93% of the global fatalities due to road crashes. In this regard, the present study aims to identify associated factors which influence fatal crashes in the context of an LMIC. Also, it aims to investigate if these associated factors are different for different road categories. The work is carried out by analysing 20,556 police-reported crash data obtained from the state of West Bengal in India. Various factors considered in analysis include roadway characteristics, vehicle characteristics, crash characteristics and human-related factors. The analysis of data using association rules mining reveals that factors associated with fatal crashes vary across different categories of roads. While causal factors on high-speed corridors, i.e. National Highways (NH) and State Highways (SH) show some similarities, such as collision with pedestrians in open area and straight sections, they are substantially different on other roads, such as hitting fixed object, involvement of two-wheeler. However, regardless of road category, speeding and absence of speed limit were found to be important associated factors in all categories of road. The findings derived from the present work may be used advantageously for formulating policy and necessary interventions to reduce fatalities.