Survival Analysis of Bus Running Time near Bus-Stop Areas

Abstract

This study proposes a quantitative approach to evaluate the effects of mixed traffic flow on bus running times (except bus dwell times) near bus-stop areas based on linear regression and survival analysis theory. Research data were collected by video cameras at four bus stops in Nanjing, China. The application of the proposed methods with field data indicated that several factors would delay bus running times, i.e., car and nonmotor volume, bus dwell time, bus berth without violation, and nonmotor violations. In addition, the effect of bus lane-changing behavior on bus running times varied from section to section. Moreover, linear and parametric survival models were also developed to estimate bus running times, and both models can capture the effect of factors. The parametric survival models have better fitness performance than the linear models. However, the predictive performance of the two models are distinct at different sections and bus stops. The findings of influential factors and the proposed models could be considered in advanced bus information systems.