Correlations among Various Parameters in Car-Following Models with NGSIM Trajectory Data

Abstract

Based on NGSIM empirical trajectory data, we focus on verifying the correlations among various parameters in car-following models. The data is preprocessed primarily by symmetric exponential moving average filter. Then, based on SPSS, we explore the preprocessed NGSIM trajectory data by plotting scatter diagram and correlation analysis. The results show that the acceleration of the following vehicle is obviously proportional to the velocity difference and acceleration difference between the following and the preceding (next preceding) vehicle, as well as the product of such velocity difference and the following vehicle’s velocity. Furthermore, we analyze the delay time by using time series correlation formula and modified R/S and V/S statistics methods.