An important part of human activities in an urban environment is their mobility behavior. Nowadays, measuring the movement of people is a fundamental activity in modern societies. First insight regarding the mobility within a region can be captured by extracting the origin–destination (OD) matrix, which specifies the travel demands between the origin and destination nodes on a network. This matrix could be on different scales including macroscopic scales, e.g., at the inter-urban level, or at microscopic scales, e.g., at the intra-urban level. In the intra-urban level, OD matrix is indicative of the movement of people between different areas of the city.

Many methods have been suggested for OD-matrix estimation, which can be classified into three main categories: survey-based methods, traffic counts, and methods based on the positioning technology. Using location-Based Social Networking (LBSN) data is a method based on the positioning technology which is raised in recent literatures as a new travel demand data source. Users of LBSN provide location-sensitive data interactively via mobile devices, including smartphones and tablets. These data has the potential to provide origin-destination movement estimation with significantly higher spatial and temporal resolution at a much lower cost in comparison with traditional methods. Data concluded from these networks are one of the modern and updated data sources that attracted researches attentions to urban management subject and let theme investigate three significant aspects of human movements and interactions including location, time and social communities.

On the other hand, various models have been proposed for the estimation of OD matrix. In this paper, two OD estimation models have been utilized in order to investigate their relative performance when using LBSN data. The first model used in this study is “radiation model” in which the number of trips is estimated with regard to population of the origin and destination location, and the total population in the circle centered at the origin with the distance between the origin and destination as the radius (excluding the origin and destination population). The second model is “population-weighted opportunities model”, in which the travel rate from origin to destination is calculated with regard to the attraction of destination. This model assumes that the attraction of a destination for a traveler is the destination’s opportunities inversely weighted by number of opportunities between the destination and the origin.

Although the two mentioned models have no adjustable parameters, they require information on variables such as population distribution and location attraction as input. In order to extract these inputs from location-based social networking data and make the proposed models compatible with this kind of data, it is essential to renew and advance the models, considering specific characters and limitations of LBSN data.

This study examines the efficiency of LBSN data in the estimation of the intra-urban OD matrix. For this, Foursquare check-ins data for Manhattan, one of the five boroughs of New York City, is collected. As each LBSN data records has a time attribute, check-ins are sorted based on time and individuals trajectories extracted using consecutive check-ins. The study area is then partitioned based on census tracts and using aggregated trajectories between these areas, the movement flow intensity between each pair of tracts is estimated by each of the mentioned models. Finally, the outputs of models are evaluated using real observations based on different criterions including distance distribution, destination travel constraints and flux. The results demonstrate the promising potential of using LBSN data for urban travel demand analysis and monitoring.