Mining crowdsourced data on bicycle safety critical events

Abstract

Cycling has become a popular transportation mode for short term trips. Due to the high exposure bicycle trips, the number of collisions and near miss events has been increasing significantly. This study explores the pattern of the bicycle-related collision or near miss events by using a unique crowdsourced dataset collected from BikeMaps.org. The dataset not only contains near miss events, which are not included in the conventional state-maintained crash databases, and it also includes the psychological impact of the event on the cyclist. The taxicab correspondence analysis (TCA) results reveal patterns for bike-related collision or near miss events and associated impact on the cyclists involved. Several factors such as inclement weather, windy condition, poor lighting conditions, wet ground, loose sand, or dirt pavement are associated with the increasing probability of the collision or near-miss events. The study indicates that collision or near miss events have a greater impact on cyclists if the events occurred when cyclists already have taken extra caution while cycling. These cyclists tend to cycle less and be more careful after these events. Interestingly, the results find that frequent cyclists are not psychologically affected by collisions occurred during recreational trips. The finding of this study could help researchers further understand bike collisions/near miss events and provide better countermeasures to mitigate the frequency of bike collisions.

Keywords

• Bikemaps.org
• Bicycle collision
• Bicycle near miss event
• Crowdsourced data
• Taxicab correspondence analysis