Undergraduate Program
Client Project Challenge
City of Evanston Bicycle Rack Distribution Analysis

The goal of this project was to provide the City of Evanston with a statistically-backed recommendation for the placement of bicycle racks in strategic areas of interest, both satisfying the needs of Evanston residents and promoting bike ridership. Our client, Catherine Hurley, coordinates Sustainable Programs for the City of Evanston. Given a redistributive surplus of racks and a total capacity of over 900 parking spaces, Catherine tasked us with creating a rack layout that would meet the demand for bikes and thereby reduce bike parking at parking meters, trees, and other unconventional and inconvenient parking locations in Evanston. We were instructed to focus our efforts outside of the downtown area, as Catherine felt that any recommended changes would be easier to implement in areas with fewer stakeholders.

Our approach primarily consisted of data collection at a sample of existing rack sites and creating a regression model, which we then applied to a new set of locations for bike racks recommended by bike enthusiasts and experts whom we surveyed. This model would test for the feasibility of placing a rack at a specific location. In order to do this, we first segmented the users into three groups, which included bicyclists who cycled for recreational, commuting, or commercial purposes. We then created four categorical variables (Recreation, Jobs, Transportation, and Commercial) in an attempt to describe each bike rack location. We found various data from the U.S. census and City of Evanston commerce webpage that helped us to numerically describe these categories.

Next, we added a binary variable indicating whether the observation was from the week or weekend; this variable was then interacted with all the initial variables so as to incorporate the effects caused by the varying of the day of the week. We found the ideal subset of predictors to be included in the model through observing the lowest p-values and narrowing the predictors down to the most sensible and statistically sound model. Finally, we applied the model to the set of locations provided by our survey, and found a list of locations that we ranked according to the greatest need for bike parking spaces.

A full report on this project is available by request to

Team: Michael Dornbusch, David Erbs, Jimmy Fang, Scott Kellert
Advisor: Professor Diego Klabjan