Demographic Information Inference from Passively Collected Data

Abstract

The growing reliance on data for transportation decision-making and research has underscored the critical importance of data equity. Ensuring fairness and justice in data representation from diverse communities is crucial to avoid biased outcomes that perpetuate transportation planning and policymaking inequities. However, passively collected big data, such as smartphone app-based data, often lacks individual-level demographic information due to privacy concerns. To address this limitation, this study focuses on smartphone app-based data and introduces a Data Generation Process (DGP) model to understand how smartphone identities are selected. Additionally, a Bayesian inference method is proposed to infer demographic information at the zone level. The method’s validation through real-world app-based data illustrates its effectiveness in providing insights into demographic variations in transportation data. By contributing to data equity in transportation, this study aims to foster a more inclusive and equitable future for transportation planning and decision-making.