Fusing Multiple Sources of Data to Understand Ride-Hailing Use

Abstract

The rise of ride-hailing services has presented a number of challenges and opportunities in the urban mobility sphere. On the one hand, they allow travelers to summon and pay for a ride through their smartphones while tracking the vehicle’s location. This helps provide mobility for many who are traditionally transportation disadvantaged and not well served by public transit. Given the convenience and pricing of these mobility-on-demand services, their tremendous growth in the past few years is not at all surprising. However, this growth comes with the risk of increased vehicular travel and reduced public transit use, increased congestion, and shifts in mobility patterns which are difficult to predict. Unfortunately, data about ride-hailing service usage are hard to find; service providers typically do not share data and traditional survey data sets include too few trips for these new modes to develop significant behavioral models. As a result, transport planners have been unable to adequately account for these services in their models and forecasting processes. In an effort to better understand the use of these services, this study employs a data fusion process to gain deeper insights about the characteristics of ride-hailing trips and their users. Trip data made publicly available by RideAustin is fused with census and parcel data to infer trip purpose, origin/destination information, and user demographics. The fused data is then used to estimate a model of frequency of ride-hailing trips by multiple purposes.