The Relationship of Built Environment and Weather with Bike Share –Evidence from the Pronto Bike Share System in Seattle

Thesis (Master's)--University of Washington, 2016-06

The purpose of this research is to identify correlations with bike station ridership for Pronto, a bike share program in Seattle. The daily number of trips and station-level trips from October 13th, 2014 to October 12th, 2015 was obtained from Pronto. Data for independent variables were from various sources. Polynomial regression models with cubic and quadratic terms are used to evaluate the relationship between weather and temporal factors on daily system-wide ridership and the effects on different types of users. Multiple linear regression models are used to investigate the effects of built environment variables on station-level ridership in 0.25-mile, 0.5-mile, and 0.75-mile scales. The models have high goodness of fit and identify a number of variables having statistically significant correlations with ridership. Temperature and wind speed are not linearly associated with daily ridership. Rain, weekends, and holidays decrease daily ridership. Different effects of weather and temporal factors on annual members and short-term pass holders are also captured in this research. Annual members, who are less affected by rainfalls than short-term pass holders, are more likely to use Pronto on weekdays while short-term pass holders tend to use Pronto on weekends and holidays. In addition, the station-level ridership is negatively associated with job density, proximity to parks, and proximity to waterfront in all three buffers. The findings will help planners and managers to predict daily ridership, optimize bike locations, and improve bike rebalance efficiency.