Online Program

Return to main conference page
Friday, February 15
Fri, Feb 15, 5:15 PM - 6:30 PM
St. James Ballroom
Poster Session 2 and Refreshments

Statistical Analysis of Large-Scale Public Transport Data (303808)

View Presentation View Presentation

*Daniel Joseph Graham, Imperial College London 

Keywords: Transportation, Big Data, Causality, Performace Modelling

In this paper we showcase the application of statistical techniques for large scale mass transportation datasets to measure system performance and identify areas for operational improvement. We present three case studies. The first merges smart card and train movement data to develop a comprehensive method to estimate the user cost of crowding in a revealed preference route choice framework. Using merged data we are able to recover fluctuations in crowding conditions on an entire metro system, including the density of standing passengers and the probability of finding a seat. The second case study evaluate the impacts of metro pricing policies on ridership via a causal inference analysis of smart card data on Honk Kong metro. The third case study applies semiparametric regression techniques to large scale automated data to benchmark and decompose journey time performance on London Underground. Our case studies demonstrate the practical usefulness for public transport operations and planning that can be gained from the application of data science techniques to large scale automated data.