The effect of cyclist’s gender on its road safety, with temporal and spatial variation

This paper is concerned with the modelling of the reported cyclist road traffic crashes by considering the gender of the cyclists. This personal attribute of the cyclist has been reported as a critical variable. However, there are very few works in the literature dealing with such a problem or undertaking modelling of this variable. There were 21 different types of variables considered for each crash, for the study area of Tyne and Wear county (north-east of England). A hybrid novel framework is proposed for the construction of road safety model. In the first step, the crash rates are determined for the study area and then evaluated with the modal share rate. This is followed by the construction of the heatmaps. In the next step, a deep learning-based crash model is constructed using historic crash data. This is followed by the statistical evaluation using Chi-square test, and Cramer’s V value. The study demonstrates that infrastructure, spatial variables, and environmental conditions affect the safety interactions of the cyclist. The research depicts a social inequity in the risk that the cyclists face; males have a higher tendency towards crash than women, and elderly women are at a relatively higher risk. However, the risk for both males and females are the highest in their youth. The safety of the cyclists varies with gender, both temporally as well as spatially. The infrastructural hazards present different level of risk to the cyclist based upon its gender, and the manner of association
of different identified variables with gender is different. The female shows a much higher
level of statistical association with the identified variable than the males. By combining
the statistical, inference, and data learning approach, we have demonstrated that a road
safety model can be constructed with significantly high accuracy and predictive power