Abstract: Public transportation agencies can obtain large amounts of information regarding timeliness, efficiency, cleanliness, ridership and other performance measures. However, these metrics are based on the interests of these agencies and do not necessarily represent the concerns of the customers. Recently, social media have become a platform for people to show their satisfaction or discontent about particular services and products (e.g., Twitter feeds, Yelp reviews, Change.org petitions). The goal of this project is to develop a tool using machine learning techniques to discover factors affecting public transit rider satisfaction using social media data. This tool is intended to reveal features of ridership that are not evident to transit agencies. For instance, a sense of community and pride are positive aspects of ridership that are not measured by traditional surveys. Specifically, sentiment analysis techniques will be utilized to classify numerous sets of rider sentiment data over a period of time and for particular locations (e.g., a Metrolink station). Each aspect will be shown as a theme in a geographic information system (GIS) layer. This online GIS-based tool can be accessed by transportation planners to determine areas of service where they can focus their resources either for the short term or long term.
Quarterly Progress Reports
Click on the links below to download any of the quarterly progress reports already finished.