Machine Learning Travel Mode Choices: Comparing the Performance of an Extreme Gradient Boosting Model with a Multinomial Logit Model

Transportation Research Record - Tập 2672 Số 47 - Trang 35-45 - 2018
Fangru Wang1, Catherine L. Ross2
1School of City and Regional Planning, Georgia Institute of Technology, Atlanta, GA
2School of City and Regional Planning and Civil and Environmental Enginering, Georgia Institute of Technology, Atlanta, GA

Tóm tắt

The multinomial logit (MNL) model and its variations have been dominating the travel mode choice modeling field for decades. Advantages of the MNL model include its elegant closed-form mathematical structure and its interpretable model estimation results based on random utility theory, while its main limitation is the strict statistical assumptions. Recent computational advancement has allowed easier application of machine learning models to travel behavior analysis, though research in this field is not thorough or conclusive. In this paper, we explore the application of the extreme gradient boosting (XGB) model to travel mode choice modeling and compare the result with an MNL model, using the Delaware Valley 2012 regional household travel survey data. The XGB model is an ensemble method based on the decision-tree algorithm and it has recently received a great deal of attention and use because of its high machine learning performance. The modeling and predicting results of the XGB model and the MNL model are compared by examining their multi-class predictive errors. We found that the XGB model has overall higher prediction accuracy than the MNL model especially when the dataset is not extremely unbalanced. The MNL model has great explanatory power and it also displays strong consistency between training and testing errors. Multiple trip characteristics, socio-demographic traits, and built-environment variables are found to be significantly associated with people’s mode choices in the region, but mode-specific travel time is found to be the most determinant factor for mode choice.

Từ khóa


Tài liệu tham khảo

10.1093/oxfordjournals.pan.a004868

10.1016/0047-2727(74)90003-6

Ben-Akiva M., 1985, Discrete Choice Analysis: Theory and Application to Travel Demand

10.1016/B978-008043360-8/50001-7

10.1016/j.trc.2010.10.004

Biagioni J. P., 2008, Presented at 88th Annual Meeting of the Transportation Research Board

10.3141/2399-01

10.1016/S0198-9715(98)00036-2

10.1016/j.trpro.2016.11.119

Zhang Y., Transportation Research Record: Journal of the Transportation Research Board, 141

10.1016/j.mcm.2006.02.002

Shukla N., 2015, International Symposium for Next Generation Infrastructure (ISNGI 2014), 215

10.3141/1718-01

10.3141/1854-06

10.1016/S1366-5545(99)00030-7

10.1016/S0968-090X(02)00021-9

10.1016/S0191-2615(96)00016-1

10.1016/S0965-8564(99)00043-9

10.2105/AJPH.93.9.1478

10.1016/j.jtrangeo.2009.07.003

10.1007/s11116-007-9136-6

10.1016/j.trb.2006.03.004

Ewing R., Transportation Research Record: Journal of the Transportation Research Board, 55

10.1016/j.trf.2011.01.006

10.1016/j.amepre.2008.01.004

10.3141/2042-01

10.1016/S0191-2615(98)00004-6

Georggi N. L., 2000, An Analysis of Long-Distance Travel Behavior of the Elderly and the Low-Income

10.1016/S0191-2615(99)00028-4

10.3141/1894-13

10.1023/A:1005253813277

10.1177/0361198105192700108

Murakami E., 1997, Daily Travel by Persons with Low Income

10.1016/S1361-9209(01)00024-4

10.1177/0013916502034002001

10.1080/01944361003766766

10.1007/s11116-011-9360-y

10.3141/1831-18

10.1080/01944360408976383

10.1214/aos/1013203451

Thông tin
Thông tin xuất bản

Transportation Research Record

Tập 2672 Số 47

35-45

Thông tin tác giả