Name
Abstract | ||
|---|---|---|
Forecasting the crowd flows in a city is crucial for public safety, traffic management and urban planning. Researchers proposed several methods to forecast the crowd flows. However, they omitted the acquisition of comprehensive flow data. Taxis' GPS trajectory data and bike sharing system data are often used in these works as the flow data. But they are not able to reflect the comprehensive crowd flows in a city, since they only contain the transitions of a specific transportation mode. In this paper, we propose to extract comprehensive crowd flows from mobile flow records (MFRs), a finegrained cellular data. We also use a Convolution Neural Network (CNN) based method to forecast crowd flows and compare it with traditional time series regression models. The experiments on a large-scale cellular dataset show that CNN based method can reduce the error by 28% to 77%.
|
Year | DOI | Venue |
|---|---|---|
2019 | 10.1145/3321408.3321579 | Proceedings of the ACM Turing Celebration Conference - China |
Keywords | Field | DocType |
DNN, cellular network, crowd flow forecasting | Computer science,Flow (psychology),Cellular network,Distributed computing | Conference |
ISBN | Citations | PageRank |
978-1-4503-7158-2 | 0 | 0.34 |
References | Authors | |
0 | 4 | |