PI: Corey Harper
Co-PI(s): N/A
University: Carnegie Mellon University
Industry partner: Healthy Ride
Transportation is a basic social and economic need, but most trips are done by car, which may not be environmentally sustainable with growing urban populations. Bike share systems have been deployed in many cities around the world and provide residents with a low-cost alternative to replace short car trips (i.e., trips within 3 miles). A key to success for shared bike systems is the effectiveness of rebalancing operations- or moving bikes from one location to another to increase utilization. However, these new services are still under development and accurately predicting demand remains an issue. The goals of this PITA proposal are to build a deployable model that: 1) mines and translates tweets into information useful for predicting bike share demand 2) incorporates Twitter messages and other relevant features into a machine learning tool to improve bike share demand prediction, and 3) conducts a proof-of-concept analysis and develops preliminary results to win an NSF Civil Infrastructure Systems proposal. This model will be developed in collaboration with Healthy Ride to assist them in making more informed rebalancing decisions, which will lead to improved economic viability by increasing overall system utilization and customer satisfaction in Pittsburgh, PA. This research is important and timely as cities look to reduce emissions and make the shift toward shared modes of transportation.