CityProphet: City-scale irregularity prediction using transit app logs
UbiComp2016 (The 2016 ACM International Joint Conference on Pervasive and Ubiquitous Computing), 2016/9
機械学習 (Machine Learning) データサイエンス (Data Science)
- Thanks to the recent popularity of GPS-enabled mobile phones, modeling people flow or population dynamics is attracting a great deal of attention. Advances in methods where regular population patterns with respect to factors such as holidays or weekdays are extracted have provided successful results in irregularity detection. With large-scale crowded events such as fireworks, it is crucial that there be enough time to take countermeasures against the irregular congestion, i.e., irregularity prediction. It remains a tough challenge to predict population from GPS trace logs with existing methods. To tackle this problem, we focus here on route search logs, since aggregation of the location-oriented queries of individual plans serves as a mirror of short-term city-scale events, in contrast to GPS mobility logs. This paper presents a brand new framework for city-scale event prediction: CityProphet. By our observation of data where the route search logs related to a future event are in most cases repeatable and accumulated in proportion as the event draws near, we are able to leverage the divergence between the above two properties to predict city-scale irregular events. We demonstrate through experiments using the transit app logs of over 370 million queries that our approach can successfully predict city-scale crowded events one week in advance.
CityProphet: City-scale irregularity prediction using transit app logs（外部サイト／External Site Link）