SENSITIVITY OF RISK INDICATORS UNDER MOTORWAY TRAFFIC REGIMES CLUSTERED BY SELF-ORGANIZING MAP

In this paper, we present a method for determining motorway traffic regimes where the sensitivity of motorway crash risk indicators could be better understood. Motorway crashes are usually severe. If the crash risk could be monitored, it would be possible to prevent crashes or at least diminish their severity. Using risk indicators could be a way to grasp the crash risk. Under different traffic regimes, there should exist typical crash risk. Hence, risk indicators would better grasp the risk if they were considered under concrete traffic regimes. Traffic situations are pre-processed by Principle Component Analysis (PCA) before being clustered by the Self-Organizing Maps (SOM) into traffic regimes. The sensitivity of risk indicators are analyzed under each traffic regime. Real traffic data for 16 months was used for the analysis.