Developing Tools for Identifying and Analyzing Amtrak’s Most Delayed Trains

Abstract

A major problem with long-distance passenger train travel on Amtrak is the constant delays. While it makes sense that a train is more likely to arrive to a station late than early, as a train always has to wait at each stop until its scheduled departure time and thus cannot easily gain time, it makes less sense how delays of up to an hour are common, or even expected, on many Amtrak trains today. For example, Amtrak’s train 5 had a mean arrival delay of 2 hours and 13 minutes and a median arrival delay of 1 hour and 26 minutes at its final station throughout 2021 and 2022. This is still clearly bad. But compared to other Amtrak trains, how bad actually is it? While there has been analysis into the delay of Amtrak services, such as the Northeast Regional or the California Zephyr, there are no public resources available for identifying and ordering individual Amtrak trains – such as train 5, train 29, or train 319 – by their lateness. This thesis aims to identify which trains on Amtrak have the greatest mean and median delays by analyzing historical Amtrak delay data and listing Amtrak trains by their calculated statistics of net magnitude of delay and delay normalized by route length within two time frames between 2021 and 2022. Further, once the most delayed trains have been identified, the thesis creates tools that analyze why these trains are as delayed as they are by visualizing the distributions of their delays, using hypothesis testing to verify their lateness, and analyzing data about the delay causes from the Federal Railroad Administration. Finally, this thesis uses the created tools to analyze the delay of selected Amtrak trains to find which trains have delay that is possible for Amtrak to fix, and which trains are fated to be delayed by causes outside of Amtrak’s control.