Lewis Lehe

• Doctor of Philosophy
• Professor (Assistant) at University of Illinois Urbana-Champaign

Using data from the American Public Transportation Association’s 2024 Public Transportation Vehicle Database, this paper estimates statistical models to explain the real prices of single-deck, rigid US transit buses in terms of their attributes. Non-electric bus prices have risen by 0.7% per year faster than inflation since 2010, while electric bus...

Discussions of minimum parking requirement policies often include maps of parking lots, which are time consuming to construct manually. Open source datasets for such parking lots are scarce, particularly for US cities. This paper introduces the idea of using Near-Infrared (NIR) channels as input and several post-processing techniques to improve the...

This paper introduces a framework that leverages Large Language Models (LLMs) to answer natural language queries about General Transit Feed Specification (GTFS) data. The framework is implemented in a chatbot called TransitGPT with open-source code. TransitGPT works by guiding LLMs to generate Python code that extracts and manipulates GTFS data rel...

This paper develops models of a bus route in which (i) stop spacing can vary; (ii) trip lengths are heterogeneous; (iii) demand is elastic; and (iv) passengers delay the bus. Since wider spacings make sufficiently long trips faster, and sufficiently short trips slower, they induce long trips and repel short trips. We explore two continuum-approxima...

This study surveys the errors in General Transit Feed Specification (GTFS) Static (Schedule) data for 632 US transit feeds. We do so using the Canonical GTFS Schedule Validator tool provided by Mobility Data, which checks feeds against GTFS standards. About 21% of GTFS feeds have at least one error. We explain what the most common errors are and pr...

The General Transit Feed Specification (GTFS) standard for publishing transit data is ubiquitous. With the advent of LLMs being used widely, this research explores the possibility of extracting transit information from GTFS through natural language instructions. To evaluate the capabilities and limitations of LLMs, we introduce two benchmarks, name...

Studies of transit dwell times suggest that the delay caused by passengers boarding and alighting rises with the number of passengers on each vehicle. This paper incorporates such a “friction effect” into an isotropic model of a transit route with elastic demand. We derive a strongly unimodal “Network Alighting Function” giving the steady-state rat...

Since 2013, US state and local governments have grappled with the question of how to tax ridehailing, and dozens of jurisdictions have adopted entirely different approaches for different reasons. This paper reviews their experiences, including information on tax designs, motivations, preemption, laws and revenues. Taxes are characterized by whether...

The General Transit Feed Specification (GTFS) standard for publishing tran sit data is ubiquitous. GTFS being tabular data, with information spread across different files, necessitates specialized tools or packages to retrieve information. Concurrently, the use of Large Language Models(LLMs) for text and informa tion retrieval is growing. The idea...

Studies of transit dwell times suggest that the delay caused by passengers boarding and alighting rises with the number of passengers on each vehicle. This paper incorporates such a "friction effect" into an isotropic model of a transit route with elastic demand. We derive a strongly unimodal "Network Alighting Function" giving the steady-state rat...

Ridesplitting – a type of ride-hailing in which riders share vehicles with other riders – has become a common travel mode in some major cities. This type of shared ride option is currently provided by transportation network companies (TNCs) such as Uber, Lyft, and Via and has attracted increasing numbers of users, particularly before the COVID-19 p...

This paper introduces a type of circular causation called Congestive Mode-Switching (CMS) that may arise when an increase in congestion penalizes transit relative to driving. If demand is also elastic, rising congestion persuades some transit riders to drive, which exacerbates congestion further, and so on. The paper explores this logic with a stat...

Ridesplitting -- a type of ride-hailing in which riders share vehicles with other riders -- has become a common travel mode in some major cities. This type of shared ride option is currently provided by transportation network companies (TNCs) such as Uber, Lyft, and Via and has attracted increasing numbers of users, particularly before the COVID-19...

The dynamic spatiotemporal characteristics of queues at urban intersections are crucial to traffic operation tasks such as signal performance measure and signal optimization. This paper addresses the high time-resolution estimation of queue profile at urban signalized intersection using Extended Kalman Filtering (EKF) with data of connected vehicle...

Transit agencies have been removing a large number of bus stops, but discussions around the bus stop spacings exhibit a lack of clarity and data for comparison. This paper proposes new terminology and concepts for statistical consideration of stop spacings, and introduces a python package and open-source database which uses General Transit Feed Spe...

This paper incorporates two novelties into steady-state, macroscopic models of street-hail taxi service: (i) heterogeneous drivers; (ii) explicit treatment of a competitive rental market for medallions (the rights to cruise for passengers). When drivers vary only by reservation wage, issuing medallions raises every driver’s take-home pay, and the s...

This paper incorporates two novelties into steady-state, macroscopic models of street-hail taxi service: (i) heterogeneous drivers; (ii) explicit treatment of a competitive market for medallions (the rights to cruise for passengers). When drivers vary only by reservation wage, the social optimum requires subsidy, the second-best does not require me...

This paper considers static models of traffic in downtowns which can produce multiple equilibria. Models with one mode of travel can have one equilibrium in the light congestion regime, and multiple in hypercongestion. Models with two modes, which differ by occupancy, can have multiple in hypercongestion and in light congestion due to mode-switchin...

This article introduces a database of bus stop spacings for 43 cities in the United States derived from GTFS files published in late 2019. Weighting each spacing by the number of times a bus traverses it, we produce distributions and summary statistics. The overall mean spacing is 313 meters. Las Vegas’ RTC has the widest mean spacing (482 m) and P...

For tolling in urban areas, the measured elasticity of demand with respect to money cost is much larger when tolls are introduced than when tolls are raised. This paper names this phenomenon “Large Elasticity at Introduction” (LEI). It has been observed in four of five cities with urban tolls: London, Singapore, Stockholm and Gothenburg. After desc...

Systems that depend on matching often exhibit scale economies, whereby increased participation leads to improved performance for all users. This paper examines the presence of such increasing returns to scale in carpool matching. Data from Scoop, a carpooling app, is used to demonstrate this phenomenon across various markets using regression. As th...

This paper considers the monetary and time costs of producing Findings (formerly Transport Findings ). After enumerating the journal’s expenses, we find the marginal monetary cost of an article is, on average, about $65, and that the journal incurs $1966 in fixed costs per year. Also, using data from a survey of Findings ’ reviewers and estimate of...

Static traffic models, in the tradition of Walters (1961), typically feature a "demand curve" giving the vehicle flow demanded for each unit travel time (inverse speed). Traditionally, the demand curve declines because people want to drive more as travel times fall. This paper proposes that the vehicle flow demanded can, instead, plausibly rise wit...

Studies of road pricing in which the Value of Time (VOT) varies among travelers suggest that road pricing benefits travelers with high VOT and hurts travelers with low VOT. This happens because, when a toll reduces congestion, only high-VOT travelers value the time saved more than the money cost. This paper uses a static traffic model with elastic...

Static traffic models, in the tradition of Walters (1961), typically feature a "demand curve" giving the vehicle flow demanded for each unit travel time (inverse speed). Typically, the demand curve declines because people want to drive more as travel times fall. This paper proposes that the vehicle flow demanded can plausibly rise with unit travel...

“Downtown congestion pricing” (DCP) is what this paper calls the practice of relieving congestion by charging tolls, or requiring the purchase of supplementary licenses, to drive in a city’s central areas. Five major cities have implemented DCP: Singapore, London, Stockholm, Milan and Gothenburg. This paper reviews the history of DCP, reporting on...

A growing consensus argues that minimum parking requirements (MPRs) make housing more expensive. This paper examines two claims from this discussion: (1) that MPRs discourage the construction of small units; (2) that the costs of building required parking are "passed on" to buyers and renters in the form of higher prices and rents. However, the mec...

This paper shows that severe congestion on streets controlled by traffic signals can be reduced by dynamically adapting the signal offsets to the prevailing density with a simple rule that keeps the signals’ green-red ratios invariant. Invariant ratios reduce a control policy’s impact on the crossing streets, so a policy can be optimized and evalua...

Currently, all downtown tolls are “access tolls,” meaning they charge for gross access to a zone, but tolls levied on distance-traveled are on the horizon. This paper shows how such tolls affect the distribution of trip lengths. A static model is presented in which travelers with potentially different trip lengths decide whether to drive into a dow...

This paper shows that severe congestion on streets controlled by traffic signals can be reduced by dynamically adapting the signal offsets to the prevailing density with a simple rule that keeps the signals’ green-red ratios invariant. Invariant ratios reduce a control policy’s impact on the crossing streets, so a policy can be optimized and evalua...

Land-use patterns are often the result of feedback effects in which one agent's decision both influences and is influenced by urban form. This paper argues that such feedback---of a positive or negative kind---could plausibly arise in the provision of off-street parking. A stylized model is used to illustrate feedback in the case of bundled parking...

This paper considers a signalized street of uniform width and blocks of various lengths. Its signals are pretimed in an arbitrary pattern, and traffic on it behaves as per the kinematic-wave/variational theory with a triangular fundamental diagram. It is shown that the long run average flow on the street when the number of cars on the street (i.e....

A growing literature exploits macroscopic theories of traffic to model congestion pricing policies in downtown zones. This study introduces trip length heterogeneity into this analysis and proposes a usage-based, time-varying congestion toll that alleviates congestion while prioritizing shorter trips. Unlike conventional trip-based tolls the scheme...