مرکز منطقه ای اطلاع رساني علوم و فناوري - Use of Logistic Regression Method for Identification of Risky Road Sections on Arsin‐Yomra Region of D10 Highway

Abstract :

Risky road section treatment is likely to be the most effective and straightforward strategy for accident reduction. A risky road section is any location that has a higher number of crashes than other similar locations as a result of local risk factors. Identification of risky road sections is very important since errors in this step may result in the inefficient use of resources for safety improvements. In this study, risky road sections were tried to be determined by using Logistic Regression (LR) technique. LR is a regression method which is used to distinguish distinct sets of observations and allocate new observations to previously defined groups. It has an important place in categoric data analysis. The study area was a 22 km long section of D10 highway, passing through Arsin and Yomra counties of Trabzon province. West‐east and east‐west directions of divided highway were handled separately and section length was selected as 500 m, therefore 44 sections were created. Traffic accident data for Arsin and Yomra were obtained by investigation of accident reports for the years 2006‐2010. Road environment properties were obtained from Road Inventory Data of General Directorate of Highways and by site investigation. At the end of the study, a model was obtained, in which 5 independent variables (horizontal alignment, vertical alignment, bridges, pedestrian crossings, and, special facilities) were used for obtaining categorical dependent variable (being risky or non risky of road section). Accuracy value of the obtained model (for this study area) was obtained as 75%.