Using CyclePhilly data to assess wrong-way riding of cyclists in Philadelphia

The increasing use of smartphones and low cost GPS have provided new sources for collecting data and using them to explain travel behavior. This study aims to use data collected from a smartphone application (CyclePhilly) to explain wrong-way riding behavior of cyclists on one-way segments to help better identify the demographic and network factors influencing the wrong-way riding decision making.

The data used in this study consist of two different sources: (a) Route trips data downloaded from the CyclePhilly Website contained trips detailed up to segment level, collected from May 2014 to April 2016 (12,202 trips by 300 unique users); and (b) Open Street Maps (OSM). Using ArcGIS, we calculate detour routes for each wrong way segment. We then built a mixed logistic regression model to identify the trip and riders' characteristics affecting wrong-way riding behavior. Next, we explore the characteristics of road facilities associated with wrong-way riding behavior.

Only 2.7% of travel distance is wrong-way, yet 42% of trips include a wrong-way segment. Commute trips have a higher chance of wrong-way riding. The longer the trips also include more wrong-way riding. Segments with higher detour ratios (ratio of distance with a detour to the wrong-way distance) are found to be associated with more wrong-way behavior. Compared to roads with no bike lane, roads with sharrow markings and buffered bike lane discourage wrong way riding.

This study proposes new methods that can be adapted to use naturalistic and probe data and analyze city-wide aberrant riders' behavior. These help planners and engineers choose between various types of bike infrastructure. Wrong-way riding is one application that can be investigated, but probe bicycle datasets provide unprecedented resolution and volume of data that will allow for more sophisticated safety and planning analyses.