Spatial-Temporal Taxi Demand Prediction Using LSTM-CNN
Authors:
- 1 August 2020
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1226-1230
Abstract
Spatial-temporal taxi demand prediction is vital for efficient planning and scheduling of taxis, which could improve overall service level of public transportation in megacities. However, previous research mainly focuses on predicting the taxi demand within certain areas, and seldom considers the inter-area demands, which is essential for the macro-level taxi scheduling. Therefore, this paper proposes an effective model for spatial-temporal inter-area taxi demand prediction through integrating Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM). CNN is adopted to extract the correlation between features and temporal closeness dependence while LSTM for fusing them in time series. The model is verified using the historical data in Haikou (China) and results show it is more accurate and stable than traditional LSTM in inter-area taxi demand prediction.Keywords
This publication has 13 references indexed in Scilit:
- CTS-LSTM: LSTM-based neural networks for correlatedtime series predictionKnowledge-Based Systems, 2020
- Short-term forecasting of passenger demand under on-demand ride services: A spatio-temporal deep learning approachTransportation Research Part C: Emerging Technologies, 2017
- Real-Time Prediction of Taxi Demand Using Recurrent Neural NetworksIEEE Transactions on Intelligent Transportation Systems, 2017
- Short-term travel time prediction by deep learning: A comparison of different LSTM-DNN models2017
- The Simpler The Better2017
- Learning Traffic as Images: A Deep Convolutional Neural Network for Large-Scale Transportation Network Speed PredictionSensors, 2017
- LSTM network: a deep learning approach for short‐term traffic forecastIET Intelligent Transport Systems, 2017
- Deep learningNature, 2015
- A bayesian dynamic linear model approach for real-time short-term freeway travel time predictionTransportation Research Part C: Emerging Technologies, 2011
- Long Short-Term MemoryNeural Computation, 1997