Choice-Based Service Region Assortment Problem: Equitable Design with Statewide Synthetic Data

Incorporating individual user preferences in statewide transportation planning is of great importance regarding revenue management and behavioral equity. However, an enduring challenge is that consistent population travel data remains scarce, particularly in underserved and rural areas. Moreover, large-scale optimization models are computationally demanding when considering stochastic travel demands in a discrete choice model (DCM) framework. These can be addressed with a combination of synthetic population data and deterministic taste coefficients. We formulate a choice-based optimization model, in which the mode share in each block group-level trip origin-destination (OD) is determined by a set of deterministic coefficients reflecting user preferences. In that case, statewide service region design becomes an assortment optimization problem with known parameters and linear constraints, which can be efficiently solved through linear or quadratic programming (depending on variant). We test the method using a hypothetical new mobility service considered for New York State. The proposed model is applied to optimize its service region with one of the three objectives: (1) maximizing the total revenue; (2) maximizing the total change of consumer surplus; (3) minimizing the disparity between disadvantaged and non-disadvantaged communities.