Modeling Taxi Demand with GPS Data from Taxis and Transit

摘要

Identifying factors that influence taxi demand is very important for understanding where and when people use taxis and how taxi demand relates to the availability and quality of transit service. This study used a large set of global positioning system (GPS) data from taxis in New York City, along with demographic, socioeconomic, and employment data to identify the factors that drive taxi demand. A technique was developed to measure and map transit accessibility based on the time required to access a transit vehicle from a specific location and time of day. Taxi data were categorized by pickups and drop-offs, and a hybrid cross-classification and regression model was developed to estimate the taxi demand across space and time. The study identified transit accessibility, population, age, education, income, and the number of jobs in each census tract as the factors with strongest explanatory power for predicting taxi demand. The study also includes a comparison of the cost of travel by taxi and transit for specific trips between Penn Station and each of the three major New York area airports. The model and analysis results show how the number of passengers traveling together in a group and the value they place on their time affect the likelihood of choosing taxi or transit for an airport access trip. A number of findings are presented in this report that are specific to New York City. However, the methods developed in this study and demonstrated in this report can be applied generally to cities around the United States and the world where similar GPS data from taxis and schedule information from transit are available.