Natural Hazards Research Summit 2024: Impacts of Weather on Travel Behavior Using Mobile Device Location Data: Deploying Two-Stage Machine Learning Models

Objective: This project aims to develop a framework for comparing various machine learning models, including advanced time-series models like Long Short-Term Memory (LSTM) networks, to predict the number of trips in transportation. Additionally, it implements a two-stage approach to first forecast wind speed and then analyze its impact on travel behavior. Key Points: • Outdoor weather significantly affects transportation, influencing system performance and passenger behavior. • Adverse weather conditions can degrade service levels, leading passengers to reconsider travel choices or avoid trips altogether. • There are gaps in existing research, particularly in studies that explore transit and travel behavior using mobile device location data across various modes of travel and distances. • Current research focuses primarily on explaining travel behavior, predicting weather conditions, or estimating ridership. Uniqueness of the Project This project is unique because it: • Utilizes mobile device location data to study travel behavior across different travel modes and distances, a relatively underexplored area. • Combines weather forecasting with transportation analysis, offering a comprehensive approach to understanding the impact of weather on travel behavior. • Implements advanced machine learning techniques, specifically LSTM networks, which are not commonly used in existing transportation studies. Target Audience: The primary audience for this project includes: • Transportation and Climate Researchers: Interested in the impact of weather on travel behavior and the application of machine learning in transportation studies. • Policy Makers: Looking to develop informed strategies to improve transportation system resilience against adverse weather. • Public Transportation Authorities: Aiming to optimize service levels and manage passenger expectations during different weather conditions.