A deep learning approach on short-term spatiotemporal distribution forecasting of dockless bike-sharing system

Dockless bike-sharing is becoming popular all over the world, and short-term spatiotemporal distribution forecasting on system state has been further enlarged due to its dynamic spatiotemporal characteristics. We employ a deep learning approach, named the convolutional long short-term memory network (conv-LSTM), to address the spatial dependences and temporal dependences. The spatiotemporal variables including number of bicycles in area, distribution uniformity, usage distribution, and time of day as a spatiotemporal sequence in which both the input and the prediction target are spatiotemporal 3D tensors within one end-to-end learning architecture. Experiments show that conv-LSTM outperforms LSTM on capturing spatiotemporal correlations.