Crash occurrence analysis is a traditional method for assessing traffic safety, yet more accurate or evident crash records may be necessary. However, unsafe traffic events such as harsh acceleration/braking instances occur more frequently and can be readily obtained. This study investigates the relationship between crash frequency and the occurrence of unsafe traffic events – harsh acceleration and braking events – utilizing smartphone app data across a network. The research aims to assess whether crashes can be predicted based on this data. Acceleration/braking events will be extracted from smartphone app data, enabling an analysis of their spatiotemporal distribution. This research explores whether the type of traffic events and their spatiotemporal resolution can enhance the prediction of crashes at specific sites such as intersections. Various regression models are developed and evaluated to determine the most accurate and reliable crash prediction models based on the combination of unsafe traffic events and spatiotemporal resolution. The anticipated findings will advocate for proactive approaches to traffic safety analysis and delineate the minimum requirements of unsafe traffic event data for such analysis.
| ID | pc549 |
| Presentation | |
| Full Text | |
| Tags |
