Allometric scaling of road accidents using social media crowd-sourced data

Abstract

Traffic accidents in Lebanon are constantly harvesting lives, dramatically changing others, and traumatizing those of their beloved ones. Due to the lack of statutory authority in charge of collecting and reporting accident related data, the Lebanese Road Accident Platform (LRAP) is proposed in this work as a real-time online platform to collect crash events from social media. LRAP allows for autonomous data collection, classification and visualization without human intervention, and aims to help the authorities in laying down the appropriate measures for traffic accidents prevention. After being in production for the last four years, the data extracted from LRAP was used to study the allometric scaling of accidents with respect to different parameters such as district area, population size per district and road network length. Such approach offers a new perspective on traffic accidents’ scaling and behavior as a living organism as cities grow. A seasonality trend analysis is also provided to analyze temporal clustering patterns in crash occurrence. © 2019 Elsevier B.V.