Equity analysis requires adequate representation in data to be able to quantify differences in a population. Certain data structures are more conducive to a fair representation than others. For example, Census data for underserved population segments can have higher measures of error, which can be problematic if such data is used for transportation analysis assuming average values are equally accurate for all segments. We derive an algorithm to construct districts to address the Modifiable Areal Unit Problem to ensure fairer representations of different population segments, using population synthesis as a use case. Upon obtaining a synthetic population, further need for equity analysis requires designing behavioral models that can capture these differences in the population without losing computational tractability in integrating with mobility service optimization models. Aggregate mode choice models are proposed that systematically capture k-modal taste heterogeneity down to Census block group OD pairs, such that choice-based minimization of income disparity in mobility service resource allocation can be done as a quadratic program. Policy implications are discussed.
Copyright. All Rights Reserved.