MULTIPLE LINEAR REGRESSION BASED APPROACH TO THE ANALYSIS OF MULTI-ANTICIPATIVE CAR-FOLLOWING BEHAVIOR

摘要

The paper empirically investigates the multi-anticipative car-following behavior using a multiple linear regression approach. Two important aspects of the multiple linear regression of a modified generalized linear multi-anticipative car-following model are addressed, e.g., multicollinearity between explanatory variables and serial correlation of time series data. Preliminary results show that a driver in stop-and-go traffic conditions is able to react to the stimuli of the first, second, and even the third leader. Moreover, it is found that a driver in normal traffic conditions only reacts to the stimuli of the first leader. Therefore, it is empirically showed that the stimuli a driver perceives may be different in different traffic condition with respect to multi-anticipative car-following behavior.