Binary Probabilistic Models for Pedestrians' Crossing Behaviour and Risk at the Free Left Turn: Delhi, India

Abstract

Free left-turn lanes are common at major signalized intersections of Delhi to create additional capacity, and to facilitate continuous flow to left-turning traffic. The present study aims at examining the pedestrians’ risk and road crossing behavior at free left-turns. Two most used binary probabilistic models, namely Logistic and Probit, were fitted to the data-set. Pedestrians cross the free left-turn when gaps are available within the traffic flow. Analysis of the data suggests that pedestrians’ waiting time prior to the crossing of free left-turn is very less. Gap size is a significant parameter. Pedestrians’ characteristics and type of conflicting vehicle do not influence their crossing behavior. Most of the pedestrians cross the free left turns with the gap size less than the adequate gap size. Both the univariate binary response Logistic and Probit regression models have been found to give similar results for the selected case.