Random Forest Models for Identifying Motorway Rear-End Crash Risks Using Disaggregate Traffic Data and Meteorological Information

This paper presents an approach to develop motorway Rear-End Crash Risk Identification Models (RECRIM) using disaggregate traffic data, meteorological data and crash database for a study site at a two-lane-per-direction section on Swiss right-hand driving motorway A1. Traffic data collected from inductive double loop detectors provide instant vehicle information such as speed, time headway, etc. We define traffic situations (TSs) characterized by 22 variables representing traffic status for 5-minute intervals. Our goal is to develop models that can separate TSs under normal conditions and TSs under traffic conditions preceding rear-end crashes using Random Forest encapsulated in TreeBagger class developed in MatLab by The MathWorks ([1]). Normal TSs for the whole year of 2005 were clustered to form groups of similar normal TSs that we call traffic regimes (TRs). Pre-crash TSs are classified into TRs so that a RECRIM for each TR is developed. Interpreting results of high performance models suggests that speed variance on the right lane and speed difference between two lanes are the two main causes of the rear-end crashes. In this paper, we also discuss about the applicability of RECRIM in a real-time framework. [1] Mathworks, T. Regression and Classification by Bagging Decision Trees. Volume, http://www.mathworks.com/access/helpdesk/help/toolbox/stats /br0gosr-1.html, Jan 22, 2010