The lane-changing process exerts a significant impact on traffic safety and urban congestion issues. Drivers’ different driving styles were considered in lane-changing characteristics analysis. Results of the questionnaires were first examined by SPSS, quantifying different driving styles via principal component analysis (PCA). Drivers were then categorized into two types: aggressive type and conservative type, using K-means clustering method. Vehicle trajectory data were collected separately for each driver type (aggressive and conservative), and fitted by polynomial models with high goodness of fit. Further, prediction models for lane-changing process were also furnished in both temporal and spatial fashion. A time-logarithmic model considering driving style factor was developed to predict lane-changing time, as well as a longitudinal distance prediction model. Results manifest high accuracy of the prediction models developed in this paper compared with the experimental data.