Geo-Tagged Social Media Data as a Proxy for Urban Mobility

January 2017

A test of whether a city's geo-tagged tweets can stand in for how people actually move.

Qian C, Kats P, Malinchik S, Hoffman M, Kettler B, Kontokosta C, Sobolevsky S · Advances in Cross-Cultural Decision Making (AHFE 2017), Springer · Read the paper →

Here the question is narrow and testable: can geo-tagged Twitter stand in for how people move around New York? We built a mobility network from where people tweet and when, checked it against the ground truth of official commuting figures, and fit the usual gravity model of how flows fall off with distance.

The Twitter network matched the official one closely — similar community structure, similar flows — and it carried something the official data lacks: a clock, the city hour by hour. My part was the network and clustering analysis, and the small web app we built to explore the flows.