Hybrid deep learning models for short-term demand forecasting of online car-hailing considering multiple factors

ABSTRACT

As an emerging force in the travel industry, online car-hailing (OCH) has been relying on digital and intelligent technologies for business innovation since its birth. Three hybrid deep learning models combining multi-factor external and internal features are proposed to predict the OCH demand. Convolutional neural network (CNN), long short-term memory (LSTM), bidirectional LSTM (BiLSTM), gated recurrent unit (GRU), and convolutional LSTM (ConvLSTM) are selected to extract features. Attention mechanisms are used to combine the global parts so that the importance of feature sequences at different times can be distinguished. The effectiveness of the proposed models considering all factors is proved by comparative experiments. Then, ablation experiments are performed to analyze the effects of attention module, external and internal factors. The results showed that the hybrid models performed better than the existing models under different factors. Various factors had different impacts on the departure and arrival flows and the hybrid models.

KEYWORDS:

• Online car-hailing demand prediction
• Multi-source factors
• deep learning
• attention mechanism
• hybrid model

Acknowledgments

This work is supported by the Program of Humanities and Social Science of Education Ministry of China (Grant No. 20YJA630008), the National Natural Science Foundation of China (Grant No. 52172331), the Natural Science Foundation of Ningbo (Grant No. 202003N4142), the Natural Science Foundation of Zhejiang Province, China (Grant No. LY22G010001), Healthy & Intelligent Kitchen Engineering Research Center of Zhejiang Province and the K.C. Wong Magna Fund in Ningbo University, China.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Natural Science Foundation of Ningbo; Healthy & Intelligent Kitchen Engineering Research Center of Zhejiang Province National Natural Science Foundation of China; K. C. Wong Magna Fund of Ningbo University; Humanities and Social Sciences Youth Foundation, Ministry of Education of China.