超越“交通荒漠”标签：面向以用户为中心的智能出行服务设计的路径诊断
Beyond the Desert Label: A Pathway Diagnostic for User-Centered Smart Mobility Service Design
浏览论文内容
中文总结 AI 辅助
本文提出一种基于路径的数据驱动诊断方法，区分交通错配与最低服务失效，并在四城市应用中揭示不同公交基线下的服务缺陷模式，以支持以用户为中心的智能出行规划。
中文摘要 AI 辅助
智慧城市出行平台日益依赖空间筛查工具来识别公共交通无法满足依赖型用户需求的社区，但单一的“交通荒漠”标签可能掩盖截然不同的用户问题：服务与集中需求之间的局部错配，或基本可用服务的缺失。这些问题需要不同的以用户为中心的应对措施。本文提出了一种基于路径、可复现、数据驱动的诊断方法，用以区分相对交通错配与最低服务失效，并报告驱动每种分类的具体服务属性（频率、运营时长、周末服务、步行可达性和目的地可达性）。该工作流程整合开放数据（GTFS、ACS、LEHD、人口普查和OpenStreetMap），利用局部莫兰指数检测空间上连贯的错配，并应用以公平为导向、聚焦弱势用户的服务失效测试。将该诊断应用于巴尔的摩、费城、纳什维尔和达拉斯，结果显示，传统公交城市以局部错配为主，而汽车导向城市则表现出更广泛的最低服务失效，且每种情况下的服务缺陷特征各不相同。通过明确服务不足标签背后的机制，该工具支持在不同公交基线水平的城市中开展更具包容性、以用户为中心的智能出行规划。
英文摘要
Smart-city mobility platforms increasingly rely on spatial screening tools to identify neighborhoods where public transit fails dependent users, but a single transit desert label can mask very different user problems: localized mismatch between service and concentrated need, or basic absence of usable service. These call for different user-centered responses. This paper introduces a pathway-based, reproducible, data-driven diagnostic that distinguishes relative transit mismatch from minimum-service failure and reports the specific service attributes (frequency, span, weekend service, walking access, and destination accessibility) driving each classification. The workflow combines open data (GTFS, ACS, LEHD, Census, and OpenStreetMap), detects spatially coherent mismatch using Local Moran's I, and applies an equity-informed service-failure test that centers vulnerable users. Applied to Baltimore, Philadelphia, Nashville, and Dallas, the diagnostic shows that legacy-transit cities are dominated by localized mismatch, while auto-oriented cities show broader minimum-service failure, with distinct service-deficit profiles in each case. By making the mechanism behind an under-service label explicit, the tool supports more inclusive, user-centered smart-mobility planning across cities with different transit baselines.
发表机构
- Syracuse University（雪城大学）
机构由 AI 辅助整理，请以论文原文为准。
