INTECIN   20395
INSTITUTO DE TECNOLOGIAS Y CIENCIAS DE LA INGENIERIA "HILARIO FERNANDEZ LONG"
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Improving human mobility prediction with geo-tagged pictures
Autor/es:
ANDRÉ PANISSON; MARIANO G. BEIRO; CIRO CATTUTO; MICHELE TIZZONI
Lugar:
Milán
Reunión:
Conferencia; NetMob 2017; 2017
Resumen:
Predicting human mobility flows at different spatial scales and aggregation levels has interesting applications in areas such as urban planning, resource allocation and disease spreading control, for example. In this sense, the vast amounts of geolocated data currently generated by mobile devices as cellphones or digital cameras provide a unique opportunity for improving our understanding of human mobility processes, with immediate benefits.In this work we analyze a set of 18 million timestamped, georeferenced pictures from Flickr, taken by 40,000 users in the U.S, which are part of the Yahoo Flickr Creative Commons 100M public dataset [1].We show that by assimilating these data into well-established mobility models as the gravity model or the radiation model we can improve their predictions significantly.