arXiv subjects

Lewis Lehe

Publications and source records attributed to Lewis Lehe.

4 recordsLinked to original sources

arXiv2024

TransitGPT: A Generative AI-based framework for interacting with GTFS data using Large Language Models

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 relevant to a query, which is then executed on a server where the GTFS feed is stored. It can accomplish a wide range of tasks, including data retrieval, calculations, and interactive visualizations, without requiring users to have extensive knowledge of GTFS or programming. The LLMs that produce the code are guided entirely by prompts, without fine-tuning or access to the actual GTFS feeds. We evaluate TransitGPT using GPT-4o and Claude-3.5-Sonnet LLMs on a benchmark dataset of 100 tasks, to demonstrate its effectiveness and versatility. The results show that TransitGPT can significantly enhance the accessibility and usability of transit data.

cs.CL↗

arXiv2023

ChatGPT for GTFS: Benchmarking LLMs on GTFS Understanding and Retrieval

The General Transit Feed Specification (GTFS) standard for publishing transit 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 information retrieval is growing. The idea of this research is to see if the current widely adopted LLMs (ChatGPT) are able to understand GTFS and retrieve information from GTFS using natural language instructions without explicitly providing information. In this research, we benchmark OpenAI's GPT-3.5-Turbo and GPT-4 LLMs which are the backbone of ChatGPT. ChatGPT demonstrates a reasonable understanding of GTFS by answering 59.7% (GPT-3.5-Turbo) and 73.3% (GPT-4) of our multiple-choice questions (MCQ) correctly. Furthermore, we evaluated the LLMs on information extraction tasks using a filtered GTFS feed containing four routes. We found that program synthesis techniques outperformed zero-shot approaches, achieving up to 93% (90%) accuracy for simple queries and 61% (41%) for complex ones using GPT-4 (GPT-3.5-Turbo).

cs.IR↗

arXiv2023

Scale Effects in Ridesplitting: A Case Study of the City of Chicago

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 pandemic. Previous findings have suggested ridesplitting can lower travel costs and even lessen congestion by reducing the number of vehicles needed to move people. Recent studies have also posited that ridesplitting should experience positive feedback mechanisms in which the quality of the service would improve with the number of users. Specifically, these systems should benefit from economies of scale and increasing returns to scale. This paper demonstrates evidence of their existence using trip data reported by TNCs to the City of Chicago between January and September 2019. Specifically, it shows that increases in the number of riders requesting or authorizing shared trips during a given time period is associated with shorter trip detours, higher rates of riders being matched together, lower costs relative to non-shared trips, and higher willingness for riders to share trips.

stat.AP↗

arXiv2022

Bus Stop Spacings Statistics: Theory and Evidence

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 Specification data to derive real-world stop spacing distributions

stat.ME↗