skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: A Machine-Learning Decision-Support Tool for Travel-Demand Modeling

Abstract

Utility maximization(UM) models are the lifeblood of virtually all travel demand models (TDM) in practice. Be it the traditional travel demand models or more advanced activity-based models, utility maximization models are used extensively to model and predict myriad travel choices such as location choice, mode choice, route choice etc. More recently machine learning (ML) models are being applied in a variety of contexts to predict choice patterns (product suggestions on Amazon, restaurant suggestions on Yelp etc.,). In the TDM arena, there has been interest in incorporating ML models where they can enhance prediction accuracy. Though there have been sporadic efforts at comparing specific utility maximization models to machine learning models, there is a need for a standard comparison tool which can evaluate ML models against UM models for a given choice context. Addressing this need, we present a tool for applying an array of models including logit, nested logit, neural network, Naive Bayes and decision tree classifiers. The tool is specifically tailored to aid in the deciding the best model for a given choice context and can be used to choose an appropriate model family or to construct a model ensemble to improve upon current modeling standards in travel demandmore » modeling. We test our proposed system on household vehicle count and work schedule targets from the 2017 National Household Travel Survey. Results demonstrate that for some variables, logit are not the most effective models, and the proposed system can aid in selecting a better model.« less

Authors:
 [1]; ORCiD logo [2]; ORCiD logo [2]
  1. University of South Alabama
  2. National Renewable Energy Laboratory (NREL), Golden, CO (United States)
Publication Date:
Research Org.:
National Renewable Energy Lab. (NREL), Golden, CO (United States)
Sponsoring Org.:
USDOE Office of Energy Efficiency and Renewable Energy (EERE), Vehicle Technologies Office (EE-3V)
OSTI Identifier:
1494741
Report Number(s):
NREL/PO-5400-72993
DOE Contract Number:  
AC36-08GO28308
Resource Type:
Conference
Resource Relation:
Conference: Presented at the Transportation Research Board (TRB) Annual Meeting 2019, 13-17 January 2019, Washington, D.C.
Country of Publication:
United States
Language:
English
Subject:
33 ADVANCED PROPULSION SYSTEMS; machine learning; decision support tool; travel demand modeling

Citation Formats

Brown, C. Scott, Garikapati, Venu, and Hou, Yi. A Machine-Learning Decision-Support Tool for Travel-Demand Modeling. United States: N. p., 2019. Web.
Brown, C. Scott, Garikapati, Venu, & Hou, Yi. A Machine-Learning Decision-Support Tool for Travel-Demand Modeling. United States.
Brown, C. Scott, Garikapati, Venu, and Hou, Yi. 2019. "A Machine-Learning Decision-Support Tool for Travel-Demand Modeling". United States. https://www.osti.gov/servlets/purl/1494741.
@article{osti_1494741,
title = {A Machine-Learning Decision-Support Tool for Travel-Demand Modeling},
author = {Brown, C. Scott and Garikapati, Venu and Hou, Yi},
abstractNote = {Utility maximization(UM) models are the lifeblood of virtually all travel demand models (TDM) in practice. Be it the traditional travel demand models or more advanced activity-based models, utility maximization models are used extensively to model and predict myriad travel choices such as location choice, mode choice, route choice etc. More recently machine learning (ML) models are being applied in a variety of contexts to predict choice patterns (product suggestions on Amazon, restaurant suggestions on Yelp etc.,). In the TDM arena, there has been interest in incorporating ML models where they can enhance prediction accuracy. Though there have been sporadic efforts at comparing specific utility maximization models to machine learning models, there is a need for a standard comparison tool which can evaluate ML models against UM models for a given choice context. Addressing this need, we present a tool for applying an array of models including logit, nested logit, neural network, Naive Bayes and decision tree classifiers. The tool is specifically tailored to aid in the deciding the best model for a given choice context and can be used to choose an appropriate model family or to construct a model ensemble to improve upon current modeling standards in travel demand modeling. We test our proposed system on household vehicle count and work schedule targets from the 2017 National Household Travel Survey. Results demonstrate that for some variables, logit are not the most effective models, and the proposed system can aid in selecting a better model.},
doi = {},
url = {https://www.osti.gov/biblio/1494741}, journal = {},
number = ,
volume = ,
place = {United States},
year = {Fri Feb 08 00:00:00 EST 2019},
month = {Fri Feb 08 00:00:00 EST 2019}
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share: