Fixed-route transit in the U.S.A. is facing stiff competition from more convenient mobility options. Microtransit—shared transportation that offers dynamic routing and scheduling to efficiently match demand—is emerging as an ally to fixed-route services. However, its positive impacts are too often constrained by the politics and economics imposed by existing transit infrastructure. This paper proposes a solution that “flips transit on its head.” By rapidly prototyping microtransit services across cities and analyzing supply-demand mismatches, it is possible to launch truly data-driven transit services. To illustrate the framework, a unique dataset generated from a year of Dallas Area Rapid Transit’s GoLink service, one of the largest on-demand microtransit services in North America, is used. Mapping and machine learning are combined to empower planners to “join the dots” when (re)designing fixed-route transit lines. It is shown that microtransit should not simply fill in the gaps left by inefficiently scheduled bus routes: by incorporating it fully into their planning processes, cities and transit agencies could dramatically reverse the fortunes of public transit.

美国的固定路线交通正面临来自更便利的出行选择的激烈竞争。微型交通——提供动态路由和调度以有效匹配需求的共享交通——正在成为固定路线服务的盟友。然而，其积极影响往往受到现有交通基础设施强加的政治和经济的限制。本文提出了一种“颠倒过境”的解决方案。通过快速构建跨城市的微型交通服务原型并分析供需失配，可以推出真正由数据驱动的交通服务。为了说明该框架，我们使用了达拉斯地区快速交通 GoLink 服务（北美最大的按需微交通服务之一）的一年生成的独特数据集。地图绘制和机器学习相结合，使规划者能够在（重新）设计固定路线公交线路时“加入点”。结果表明，微型交通不应简单地填补低效公交路线留下的空白：通过将其完全纳入其规划过程，城市和交通机构可以极大地扭转公共交通的命运。