- Web site:
- Contact:  Hans Engler, <firstname.lastname@example.org>
- Short Description:  Presentation on Analysis of Capital Bikeshare Transaction Data
Nathan Davis, Ryan McMillin, Michael Slattery (graduate students) This presentation is about a data analysis project that was carried out by the presenters during the Spring semester. Since October 2010, Capital Bikeshare has been offering rental bicycles throughout the District of Columbia and in Arlington at stations such as the one outside Georgetown University's Healy Gate. In the first fifteen months, there were more than 1.3 million individual rentals. The data for each individual transaction are available to the public. As a spreadsheet, this is a file of about 200 MB, comparable to daily transaction data for a major retail company. By analyzing the data, it is possible to find out which routes are most popular, how the rental business changes by hour, by weekday, or by season, which stations are often in danger of running out of bikes or of being full, and much more. In this talk, the speakers will report on a project in which the publicly available data set was analyzed, some interesting and some less obvious information in the data was found and displayed, and stochastic models for the entire system and for some of its parts were developed and simulated numerically. The project also identified some questions that cannot be answered because even 1.3 million transactions are not enough. Among others, the following questions will be addressed: How reliable are the data as a whole? How do the traffic patterns at various stations differ? How can one forecast the traffic at an individual station or in the system as a whole? How should one redistribute the bikes over night so that most station have bikes and free slots available throughout most of the following day?