Journal of Korean Society of Transportation ISSN:1229-1366(Print) 2234-4217(Online) 대한교통학회지

In this study, traffic accident models are developed to predict traffic accident rate based on the congestion level in the stage of pre-feasibility study. 8,195 of traffic accident data was collected for the last 3 years (2018-2020) in the length of 1,805km which consist of the nine routes and 220 interchange sections of expressways. Traffic accident rates were divided by the length of the section in terms of accident frequency, fatalities, and injured respectively. Traffic congestion index was set to daily V/C. The relationship between the congestion level and traffic accident rate seems to be positive linear pattern increasing traffic accident rate while congestion level is more higher. This result comes from scatter plots and box plots as well. It was motivated to adopt linear regression modeling. In simple regression model with V/C as a independent variable, modeling without intercept is the best in terms of R². But both models for fatalities and injured were the best when the intercept was set to 0 and accident rates was transformed by logarithm. A Model for accident is satisfied having 0.75 of R² but each model for injured and fatalities has 0.58, 0.51 of R² which is relatively low but acceptable value. The study has value confirming that there is a virtuous cycle effect that increasing highway capacity through highway project is affected to decrease traffic accident as well as reduce congestion level. The future study will focus on benefit estimation technique from reducing traffic accident using the developed models aiming to improving the guideline for preliminary feasibility study.