Lateral shifting behavior of vehicles at sharp horizontal curves on hilly highways: insights from binomial logistic regression

The presence of a hill on the inner side and a cliff on the outer side of a sharp horizontal curve on a hilly highway creates a hazardous environment which is prone to severe lane-departure crashes, primarily due to frequent lateral shifts of vehicles from their ideal path. This study aims to identify the factors influencing such lateral shifting and to develop predictive models under varying operating conditions. The analysis classifies lateral shifts as either toward the centerline or the edge line of the carriageway, using binomial logistic regression. The accuracy and performance of the developed models were evaluated through a confusion matrix and multiple performance metrics. Sensitivity analysis further revealed that an increase in the subject vehicle’s speed results in a tendency to shift toward the edge line on inner lanes and toward the centerline on outer lanes. Similarly, an increase in the size of the subject and preceding vehicles raises the probability of a shift toward the centerline. The empirical models presented in this study offer a practical tool for predicting lateral vehicle shifts, which is a critical factor in evaluating traffic safety at sharp horizontal curves.