Development of a binary logistic lane change model and its validation using empirical freeway data

Abstract

An effective macroscopic lane change (LC) model is required to facilitate active and dynamic lane management to develop cell-based multilane macroscopic traffic models. Existing logistic regression LC models do not undertake model classification of lane change; do not consider performance measures in the validation of field data and ignore movement between lanes during time-varying traffic. Models that consider the direction of LC are, however, biased in their prediction of left LC (LLC) and right LC (RLC) direction. This study proposed a simplified macroscopic cell-based binary logistic LC (BLLC) model describing the LC probability using fewer explanatory variables; in this model, the direction of LC is considered as a new variable. Considering the model performance measures, the results show that there exists substantial difference in LC behaviour in both directions. The present model also achieved a smaller difference in the percentage of accurate prediction (0.9%) between the LLC and RLC.

KEYWORDS:

• Macroscopic lane change model
• lane change direction
• Next Generation Simulation (NGSIM) data
• probability of lane change
• binary logistic regression
• cell based

Acknowledgements

This work was supported through NGSIM data provided by the Federal Highway Administration (FHWA) of the U.S. Department of Transportation. The authors would like to thank Kong Xin Ying, Belinda Ng and Clement Song Hua Ong for assisting in the work described in this paper. Appreciation also goes to Editage (www.editage.com) and Scribendi for English language editing.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Christina Ng http://orcid.org/0000-0002-4978-8359

Susilawati Susilawati http://orcid.org/0000-0003-1831-6743

Md Abdus Samad Kamal http://orcid.org/0000-0003-3150-0510

Additional information

Funding

This research has been made possible through financial contributions from the Ministry of Higher Education Malaysia (MOHE) under the Fundamental Research Grant Scheme (FRGS) [Project code FRGS/1/2015/TKO8/MUSM/03/1].