A hybrid integrated deep learning model for the prediction of citywide spatio-temporal flow volumes

ABSTRACT

The spatio-temporal residual network (ST-ResNet) leverages the power of deep learning (DL) for predicting the volume of citywide spatio-temporal flows. However, this model, neglects the dynamic dependency of the input flows in the temporal dimension, which affects what spatio-temporal features may be captured in the result. This study introduces a long short-term memory (LSTM) neural network into the ST-ResNet to form a hybrid integrated-DL model to predict the volumes of citywide spatio-temporal flows (called HIDLST). The new model can dynamically learn the temporal dependency among flows via the feedback connection in the LSTM to improve accurate captures of spatio-temporal features in the flows. We test the HIDLST model by predicting the volumes of citywide taxi flows in Beijing, China. We tune the hyperparameters of the HIDLST model to optimize the prediction accuracy. A comparative study shows that the proposed model consistently outperforms ST-ResNet and several other typical DL-based models on prediction accuracy. Furthermore, we discuss the distribution of prediction errors and the contributions of the different spatio-temporal patterns.

KEYWORDS:

• Spatio-temporal flow volume
• prediction
• deep learning
• LSTM
• ResNet

Acknowledgments

The authors thank Prof. May Yuan, Prof. Grant McKenzie, and the anonymous reviewers for their insightful comments. The authors thank Dr. Junbo Zhang and M.S. Jie Li for providing the information of the experiment data.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work is supported by the Science and Technology Project of Qingdao under Grant number [16-6-2-61-NSH]; The first author’s joint Ph.D. research and the fifth author’s Ph.D. research are funded by the China Scholarship Council (CSC). The CSC is a non-profit institution with legal person status affiliated with the Ministry of Education in China.

Notes on contributors

Yibin Ren

Yibin Ren is a postdoctoral fellow at the Institute of Oceanography, Chinese Academy of Science. He visited the SpaceTimeLab for Big Data Analytics, University College London during Oct. 2017-Oct. 2018 as a jointly PhD student. His research interests include spatio-temporal data modelling and AI-aided information mining for ocean big data.

Huanfa Chen

Huanfa Chen is a teaching fellow in Spatial Data Science at Centre for Advanced Spatial Analysis, University College London. His research interest includes Geocomputation, spatial optimisation, space-time modelling and agent-based simulation with applications in crime, transport and social media.

Yong Han

Yong Han is a professor in Ocean University of China. His research interests include urban GIS,  spatio-temporal data mining and VRGIS.

Tao Cheng

Tao Cheng is a professor in Geoinformatics at Department of Civil, Environmental and Geomatic Engineering, University College London. She is the Founder and Director of SpaceTimeLab for Big Data Analytics. Her research interests include network complexity, Geocomputation, space-time analytics and Big data mining (modelling, prediction, clustering, visualisation and simulation) with applications in transport, crime, health, social media, and natural hazards.

Yang Zhang

Yang Zhang is a Ph.D. student with SpaceTimeLab for Big Data Analytics, University College London. Her research interests include spatio-temporal data mining, deep learning, urban com- puting and geo-computation with applications in transport, crime and social media.

Ge Chen

Ge Chen is a  professor in Ocean University of China. His research interests include satellite ocean remote sensing, ocean big data mining and ocean GIS.