: 29-36
Received: October 15, 2020
Accepted: October 25, 2020
Lviv Polytechnic National University, Software Department
Lviv Polytechnic National University, Lviv, Ukraine

The peculiarities of neural network training for forecasting taxi passenger demand using graphics processing units are considered, which allowed to speed up the training procedure for different sets of input data, hardware configurations, and its power. It has been found that taxi services are becoming more accessible to a wide range of people. The most important task for any transportation company and taxi driver is to minimize the waiting time for new orders and to minimize the distance from drivers to passengers on order receiving. Understanding and assessing the geographical passenger demand that depends on many factors is crucial to achieve this goal. This paper describes an example of neural network training for predicting taxi passenger demand. It shows the importance of a large input dataset for the accuracy of the neural network. Since the training of a neural network is a lengthy process, parallel training was used to speed up the training.

The neural network for forecasting taxi passenger demand was trained using different hardware configurations, such as one CPU, one GPU, and two GPUs. The training times of one epoch were compared along with these configurations. The impact of different hardware configurations on training time was analyzed in this work. The network was trained using a dataset containing 4.5  million trips within one city. The results of this study show that the training with GPU accelerators doesn't necessarily improve the training time. The training time depends on many factors, such as input dataset size, splitting of the entire dataset into smaller subsets, as well as hardware and power characteristics.


  1. Biao Leng, Heng Du, Jianyuan Wang, Li Li, & Zhang Xiong. (2016). Analysis of Taxi Drivers Behaviors Within a Battle Between Two Taxi Apps. IEEE Transactions on Intelligent Transportation Systems, 17(1), 296–300.
  2. Bruce Schaller. (2005). A regression model of the number of taxicabs in US cities. Journal of Public Transportation, 8(5), 4–11.
  3. Dhiraj, K. (2019). 10 reasons why PyTorch is the deep learning framework of the future. Retrieved from:
  4. Dipanjan Sarkar, Raghav Bali, & Tushar Sharma. (2018). Practical Machine Learning with Python. Springer Science+Business Media. New York.
  5. Du, K.-L., & Swamy, M.N.s. (2014). Multilayer Perceptrons: Architecture and Error Backpropagation. Neural Networks and Statistical Learning, pp. 83–126.
  6. Fei Miao, Shuo Han, Shan Lin, Qian Wang, John A. Stankovic, Abdeltawab Hendawi, Desheng Zhang, Tain He, & George J. Pappas. (2019). Data-Driven Robust Taxi Dispatch Under Demand Uncertainties. IEEE Transactions on Control Systems Technology, 17(1), 175–191.
  7. Firmino, P., de Mattos, Neto P., & Ferreira, T. (2014). Correcting and combining time series forecasters. Neural Networks, 50, 1–11.
  8. Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2016). Region-Based Convolutional Networks for Accurate Object Detection and Segmentation. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 38(1), 142–158.
  9. Grossberg, S. Z. (2010). Neural Networks and Natural Intelligence. Cambridge, MA: MIT Press, 651 p.
  10. Haykin, S. (2008). Neural Networks and Learning Machines. New Jersey: Prentice Hall, 936 p.
  11. Jason Dsouza. (2020). What is a GPU and do you need one in Deep Learning? Retrieved from:
  12. John Grinberg, Arzav Jain, & Arzav Vivek (2014). Predicting Taxi Pickups in New York City. Retrieved from:
  13. Jun Xu, Rouhollah Rahmatizadeh, Ladislau Bölöni, & Damla Turgut. (2018). Real-Time Prediction of Taxi Demand Using Recurrent Neural Networks. IEEE Transaction on Intelligent transport system, 19(8), 2572–2581.
  14. Kennedy, R. K., Khoshgoftaar, T. M., Villanustre, F., & Humphrey, T. (2019). A parallel and distributed stochastic gradient descent implementation using commodity clusters. Journal of Big Data, 6(1), 16.
  15. Kiani, K. (2005). Detecting business cycle asymmetries using artificial neural networks and time series models. Computational Economics, 26(1), 65–89.
  16. Kim, Yoon. (2014). Convolutional neural networks for sentence classification. IEMNLP, 1746–1751.
  17. Kingma, D. P., & Ba, J. (2014). Adam: A method for stochastic optimization. arXiv – preprint arXiv: 1412.6980.
  18. Krizhevsky Alex, Sutskever Ilya, Hinton Geoffrey E. (2012). Imagenet classification with deep convolutional neural networks. NIPS, 1106–1114.
  19. Krizhevsky, A. (2014). One weird trick for parallelizing convolutional neural networks. CoRR, abs/1404.5997.
  20. Lam, M. (2004). Neural network techniques for financial performance prediction: integrating fundamental and technical analysis. Decision Support Systems, 37(4), 567–581.
  21. Li, J., Nicolae, B., Wozniak, J., & Bosilca, G. (2019). Understanding scalability and fine-grain parallelism of synchronous data parallel training. IEEE/ACM Workshop – Machine Learning in High Performance Computing Environments (MLHPC) IEEE, pp. 1–8.
  22. Lopatko, O., & Mykytyn, I. (2016). Neural networks as the means of forecasting the temperature value of a transient process. Measuring Equipment and Metrology, 77, 65–69.
  23. Luis Moreira-Matias, et al. (2012). A predictive model for the passenger demand on a taxi network. International IEEE Conference on. IEEE, 15, 1014–1019.
  24. Naoto Mukai, & Naoto Yoden. (2012). Taxi Demand Forecasting Based on Taxi Probe Data by Neural Network. Intelligent Interactive Multimedia: Systems and Services. Ed. by Toyohide Watanabe et al. Smart Innovation, Systems and Technologies 14. Springer Berlin Heidelberg, pp. 589–597.
  25. Nicholas Jing Yuan, Yu Zheng, Liuhang Zhang, & Xing Xie. (2013). T-Finder: A Recommender System for Finding Passengers and Vacant Taxis. IEEE Transactions on Knowledge and Data Engineering, 25(10), 2390–2403.
  26. Önder, E., Fɪrat, B., & Hepsen, A. (2013). Forecasting Macroeconomic Variables using Artificial Neural Network and Traditional Smoothing Techniques. Journal of Applied Finance & Banking, 3(4), 73–104.
  27. Pal, S., Ebrahimi, E., Zulfiqar, A., Fu, Y., Zhang, V., Migacz, S., Nellans, D., & Gupta, P. (2019). Optimizing multi-gpu parallelization strategies for deep learning training. EEE Micro, 39(5), 91–101.
  28. PyTorch. (2020). PyTorch documentation. Retrieved from:
  29. Ren, S., He, K., Girshick, R., & Sun, J. (2017). Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks. In: IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), 1137–1149.
  30. Simonyan, K., & Zisserman, A. (2014). Very Deep Convolutional Networks for Large–Scale Image Recognition. CoRR, abs/1409.1556.
  31. YouTube. (2020). Consumer assessment of taxi services in large cities. Retrieved from: [In Ukrainian].
  32. Zhang Xiang, Zhao Junbo, LeCun Yann. (2015). Character-level convolutional networks for text classification. In: Advances in Neural Information Processing Systems, pp. 649–657.