A New Breakthrough in Travel Behavior Modeling Using Deep Learning: A High-Accuracy Prediction Method Based on a CNN

Xuli Wen and Xin Chen ()
Additional contact information
Xuli Wen: School of Civil and Transportation Engineering, Southeast University Chengxian College, Nanjing 210088, China
Xin Chen: School of Management Science and Engineering, Nanjing University of Finance and Economics, Nanjing 210023, China

Sustainability, 2025, vol. 17, issue 2, 1-14

Abstract: Accurately predicting travel mode choice is crucial for effective transportation planning and policymaking. While traditional approaches rely on discrete choice models, recent advancements in machine learning offer promising alternatives. This study proposes a novel convolutional neural network (CNN) architecture optimized using orthogonal experimental design to predict travel mode choice. Using the SwissMetro dataset, which represents a specific intercity travel context in Switzerland, we evaluate our CNN model’s performance and compare it with traditional machine learning algorithms and previous studies. The key innovations of our study include: (1) an optimized CNN architecture designed to capture complex patterns in travel behavior data, and (2) the application of orthogonal experimental design to efficiently identify optimal hyperparameter settings. The results demonstrate that the proposed CNN model significantly outperforms logit models, support vector machines, random forests, gradient boosting, and even state-of-the-art techniques combining discrete choice models with neural networks. The optimized CNN achieves a remarkable 95% accuracy, surpassing the best-performing benchmarks by 14–25%. The proposed methodology offers a powerful tool for understanding travel behavior, improving travel demand forecasting, and informing transportation planning decisions. Our findings contribute to the growing body of literature on machine learning applications in transportation and pave the way for further advancements in this field.

Keywords: travel mode choice prediction; convolutional neural networks; machine learning; orthogonal experimental design; transportation planning (search for similar items in EconPapers)
JEL-codes: O13 Q Q0 Q2 Q3 Q5 Q56 (search for similar items in EconPapers)
Date: 2025
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.mdpi.com/2071-1050/17/2/738/pdf (application/pdf)
https://www.mdpi.com/2071-1050/17/2/738/ (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:gam:jsusta:v:17:y:2025:i:2:p:738-:d:1569948

Access Statistics for this article

Sustainability is currently edited by Ms. Alexandra Wu

More articles in Sustainability from MDPI
Bibliographic data for series maintained by MDPI Indexing Manager ().