涩里番

Sentao Miao

Leeds School of Business, University of Colorado

Data-Driven Facility Layout for Enhancing Walk Appeal in Walkable Neighborhoods

Date: Friday, May 1, 2026
Time: 11:00 AM - 12:00 PM
Location: Armstrong Room 375

Abstract

In response to the challenges posed by car-centric urban development, there is growing global advocacy for walkable neighborhoods that promote public health and urban sustainability. To support the design of such neighborhoods, this paper studies the optimal layout of facilities (e.g., retail and service establishments) to enhance walk appeal. Unlike conventional designs that rely primarily on distance metrics, our approach focuses on pedestrian choices. We use a latent class logit model to estimate how facility layouts influence individuals鈥� propensity to walk, accounting for heterogeneous travel purposes. This model is calibrated with a city-wide dataset from our industry collaborator, a leading Chinese map service provider. Building on this model, we develop a choice-based optimization framework to determine facility layouts that maximize the aggregate walking probability of residents. We first analyze its structural properties to derive analytical insights into optimal layout patterns. To find optimal layouts for this NP-hard problem, we reformulate it as an exponential cone program. To address the computational challenges this poses for large-scale applications, we further develop efficient greedy algorithms and establish their performance guarantees via submodularity ratio analysis. A case study in Beijing shows that our approach significantly improves walk appeal compared to traditional methods, with minimal impact on facility accessibility. Our results also reveal a 鈥渨isdom of crowds鈥� effect in existing layouts and emphasize the importance of neighborhood context in shaping optimal design patterns. Specifically, we find that dispersed layouts are preferable in already walkable environments, while more challenging walking conditions benefit from spatially concentrated facilities.