Using Small Area Estimation to Produce Reliable Transportation Statistics: The Case of Household Trips Estimation at the Census Tract Level

• Published in: Data science for Transportation 2024-12, Vol.6 (3), Article 21
• Main Authors: Al-Khasawneh, Mohammad B., Cirillo, Cinzia
• Format: Article
• Language: English
• Subjects: Computational Intelligence; Data Mining and Knowledge Discovery; Engineering; Transportation Technology and Traffic Engineering
• ISSN: 2948-135X; 2948-1368
• DOI: 10.1007/s42421-024-00105-1

This paper proposes Small Area Estimation (SAE) methods on linked datasets to generate reliable transportation statistics in cases where data on travel behavior are limited or missing. Specifically, household person trips are estimated at the census tract by linking data from the Regional Travel Survey (RTS), the American Community Survey (ACS), and US Census 2020 data. The proposed SAE modeling framework integrates direct and synthetic estimations to produce accurate statistics. Several small-area estimation techniques have been employed, including regression-based models and population synthesis for areas with zero samples, as well as the Fay–Herriot model for areas with small samples. For the regression-based models, we assessed several models, including linear, Poisson, negative binomial, and random forest models, using cross-validation analysis. The Fay–Herriot method is also applied to improve estimation precision by combining direct and synthetic estimation approaches. Results showed the proposed methodology’s effectiveness in generating reliable estimates in both cases of missing or limited samples. The research highlights the potential of SAE methods in enhancing transportation analysis by integrating diverse datasets and reducing the survey data collection burden. These findings have practical implications for researchers, policymakers, and transportation planners seeking reliable estimates for smaller domains and subgroups using existing data sources.