Identification of Bicycle Demand from Online Routing Requests

Abstract Governments at all levels aim to increase cycling and walking within the mix of transportation modes. Accurate estimation of existing and potential bicycle trips on different parts of a road network is necessary to determine for which segments an investment in improved bicycling infrastructure is most effective. This paper introduces the novel idea to estimate bicycle demand on road segments based on logged trip requests that users submitted to a Web based bicycle trip planner. As a first step in this research direction, this paper assesses the general suitability of logged trip data for modeling cycling demand. More specifically, this study analyzes logged trip origins and destinations from user requests collected over a one-year period. The requests were submitted to an online bicycle trip planner developed for Broward County, Florida. The study then compares (a) point positions of logged trip origins with origins of bicycle commute trips obtained from census data, and (b) trip length distributions of logged trips with trip lengths obtained from observed bicycling trips in street networks. Several basic spatial and temporal filters are introduced and applied on the logged trip data to identify requests that most likely represent an actual trip and therefore provide a potential resource to predict bicycle demand