An Empirical Study on the Lane-Change Duration of Naturalistic Driving Based on Multiple Linear Regression Model

In China, where the driving environment is extremely complicated, lane-change scene occurs frequently and its impact is prominent. Analysis of lane changing driving behavior can not only provide a certain theoretical support for traffic managers, but also provide a reference basis for the development of advanced driver assistance systems. In view of the current Chinese research gaps in the model of lane-change duration, this article focuses on the lane-change duration. Firstly, using mathematical statistics to describe the general situation of lane changing behavior characteristics in common driving scenarios; then using simple univariate linear regression method to analyze the relationship between each parameter and the lane changing duration; and then, establishing a multivariate linear regression model including all independent variables, and use forward stepwise regression to screen the model parameters. There is a certain degree of improvement after the model is optimized to acquire higher determination coefficient, the overall linear significance, and the significance level of some variables. Finally, the sklearn library in python is called to train, tune and evaluate the model.