01518698

Component

As road congestion gets exacerbated in most metropolitan areas, many transportation policies and planning strategies try to nudge travelers to get off the road and use other sustainable modes. In order to better analyze these planning/policy strategies, there is imperative need in multimodal analysis and accurately modeling travelers’ mode switching behavior. In this paper, a popular artificial intelligence approach, Decision Tree, is used to explore the underlying rules of travelers’ switching decision between two modes under proposed framework of dynamic mode searching and switching. The 2007/2008 TPB Household Travel Survey data is used to calibrate and validate the decision tree models. An effective and practical method for mode switching decision tree induction is proposed. Loss matrix is introduced to handle the class imbalance issues. Important factors and their relative importance are analyzed through the information gain and the feature selection. A total number of six mode switching models between each two modes are trained with a high accuracy. Through comparison with Logit models, the improved prediction ability of the decision tree models has been demonstrated.

This paper was sponsored by TRB committee ABJ70 Artificial Intelligence and Advanced Computing Applications.

01503729

14-4067

English

20p

2014

Transportation Research Board 93rd Annual Meeting

Digital/other

Figures; References; Tables

Artificial intelligence; Choice models; Data mining; Decision trees; Modal shift; Mode choice; Travel behavior

Highways; Passenger Transportation; Planning and Forecasting; Public Transportation; I72: Traffic and Transport Planning

Transportation Research Board Annual Meeting 2014 Paper #14-4067

TRIS, TRB, ATRI

Jan 27 2014 3:24PM