01593384

Component

When both older data with large number of observations and recent data with small number of observations are available for travel demand forecasting, researchers must decide if they utilise either model updating methods (if so, which model updating method), where both the older and recent data are utilised, or only the smaller recent data. However, it has not been uncovered in which case, especially how many samples are collected from each of two points in time, which modelling approach should be selected for travel demand forecasting. This study investigates this issue in a context of commuting mode choice behaviours by utilising repeated cross-sectional data collected in Nagoya, Japan. Bootstrapping is adopted to have insights with statistical meaning. Findings are as follows: (1) when enough samples are collected from the recent time point, use of only the recent data is recommended; (2) when enough samples are not collected from the recent time point, transfer scaling, joint context estimation, and combined transfer estimation are recommended based on criteria defined by the differences in contexts between time points and the sample size from older time point.

This paper was sponsored by TRB committee ADB40 Standing Committee on Transportation Demand Forecasting. Alternate title: Criteria for Selecting Model Updating Methods for Better Temporal Transferability

01584066

16-4843

English

18p

2016

Transportation Research Board 95th Annual Meeting

Digital/other

Figures; References; Tables

Data analysis; Forecasting; Travel demand

Nagoya (Japan)

Planning and Forecasting; Transportation (General)

Transportation Research Board Annual Meeting 2016 Paper #16-4843

TRIS, TRB, ATRI

Jan 12 2016 6:07PM