Evaluation of Equivalent Property Damage Only (EPDO) weight sets for hotspot identification: A case study on Korean expressways

Abstract

The Equivalent Property Damage Only (EPDO) method, which assigns weights to different crash severities: fatal, severe injury, minor injury, and Property Damage Only (PDO) crashes, is widely used for hotspot identification. Determining appropriate EPDO weight sets is crucial, since incorrectly specified weight sets can misidentify hotspots and undermine safety interventions. However, existing EPDO weight sets are largely crash-cost-based or internally determined ratios, lacking empirical validation. To fill this gap, this study proposes a data-driven framework that systematically generates and evaluates EPDO weight sets using Korean expressway crash data. For multiple EPDO weight sets, we identify the top 1% to 10% of hotspots for each weight set using these data. The performance of the EPDO weight sets is evaluated by examining the numbers of (i) fatal, (ii) non-PDO (i.e., including fatal and injury crashes), and (iii) total crashes that occurred at those hotspots in the subsequent year. Based on these three criteria, we derive Pareto EPDO weight sets. The results indicate that assigning higher weights to non-PDO compared to PDO crashes, using weights smaller than typical cost-based values, leads to more accurate identification of hotspots in terms of both fatal and casualty crashes. Conversely, assigning identical weights across severities best predicts total crashes. This framework enables practitioners and policymakers to recalibrate EPDO weight sets to local conditions, improving the allocation of limited safety resources.