Urban Origin-Destination Travel Demand Analysis Using Location-based Social Networking (LBSN) Data

This research investigates the feasibility of using Location-based Social Networking (LBSN) data to estimated Origin-Destination (OD) matrix. The LBSN is a location-sensitive service interactively carried out by users to share their locations with their friends by "check-in" via mobile applications on a smartphone or tablet. With its increase popularity and sophistication, the LBSN data have emerged as a new data source for studying urban travel demand. Comparing with traditional OD estimation method such as survey based or traffic count based methods, LBSN data has the potential to provide OD estimation with much higher temporal resolution at much lower cost.;The study uses the check-in data in the Chicago CBD area and the Austin area available through the leading LBSN provider, Foursquare. A combined non-parametric cluster and regression model is introduced to establish the relationship between check-in counts and the trip production and attraction. Then, a modified gravity model based two-regime trip distribution method is proposed to estimate the OD matrix based on the estimated trip production and attraction, and the singly-constrained trip balancing method and doubly-constrained trip balancing method are presented. The proposed methods are applied to estimate daily OD matrices for multiple trip purposes such as the non-commuting trips, home-based work trips and home-based retail trips, as well as within-day dynamic OD matrices for general trips. The methods are evaluated against the ground truth OD data from CMAP (Chicago Metropolitan Agency for Planning) and CAMPO (Capital Area Metropolitan Planning Organization). The results illustrate the promising potential of using LBSN data to monitor long-term travel demand trend changes and dynamic travel demand patterns.