交通運輸是全球溫室氣體排放的第二大來源，其中陸上交通佔比高達 75%，為主要成因之一。為了有效減緩碳排放並因應現代人的出行習慣，除了積極推動大眾運輸的使用外，亦鼓勵民眾在使用叫車服務時選擇共乘，以降低單一車輛的使用頻率。然而，與陌生人共乘可能帶來繞路與心理不適等負面體驗，導致共乘接受度偏低。為此，叫車平台通常透過提供經濟折扣作為補償，期望提升乘客的共乘意願，但根據目前統計觀察整體共乘率仍相對偏低。

本論文提出一種基於行程特徵的兩階段式學習框架，用以預測乘客的共乘意願並推估其對折扣的接受程度分布。研究目標在於辨識具有高共乘潛力的乘客，作為折扣投放的主要對象，藉此提升整體共乘率。

該框架包含兩個階段：第一階段為半監督式分類模型，負責識別出具有高共乘潛力的叫車行程；第二階段則針對上述乘客，透過截尾迴歸模型預測其對折扣的可接受範圍。我們在目前最大規模的公開叫車資料集上進行模擬實驗，實驗結果顯示，本方法能有效提升共乘比例，展現其在實務應用中的潛力。

Transportation is the second largest source of global greenhouse gas emissions, with road transport accounting for up to 75% of these emissions, making it a primary contributor. To effectively reduce carbon emissions while accommodating modern travel habits, in addition to actively promoting the use of public transportation, individuals are also encouraged to choose ridesharing services to reduce the frequency of single-occupancy vehicle usage. However, sharing rides with strangers may bring negative experiences such as detours and psychological discomfort, leading to a generally low acceptance rate. To address this, ride-hailing platforms often offer monetary discounts as compensation in an effort to increase ridesharing willingness, though the overall ridesharing rate remains relatively low.

This paper proposes a two-stage learning framework based on trip features to predict passengers’ willingness to share rides and estimate the distribution of their discount acceptance levels. The objective is to identify passengers with high ridesharing potential as the primary targets for discount allocation, thereby improving the overall ridesharing rate.

The proposed framework consists of two stages. The first stage uses a semi-supervised classification model to identify ride-hailing trips with high ridesharing potential. The second stage applies a censored regression model to estimate the range of acceptable discounts for those high-willingness passengers. We conduct simulation experiments on the largest publicly available ride-hailing dataset, and the results demonstrate that our method can effectively increase the ridesharing rate, showcasing its potential for real-world deployment.

[1] New York City Taxi and Limousine Commission, “TLC Trip Record Data,” https://www.nyc.gov/site/tlc/about/tlc-trip-record-data.page, 2025.
[2] D. M. Storch, M. Timme, and M. Schr¨oder, “Incentive-driven transition to high ridesharing adoption,” Nature Communications, vol. 12, p. 3003, 2021.
[3] M. Dessouky et al., “Real-time large-scale ridesharing with flexible meeting points,” UC Berkeley Transportation Research Report, Tech. Rep., 2025.
[4] B. Han, H. Lee, and S. Martin, “Real-time rideshare driver supply values using online reinforcement learning,” in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’22), 2022.
[5] L. Tang, Z. Liu, Y. Zhao, Z. Duan, and J. Jia, “Efficient ridesharing framework for ridematching via heterogeneous network embedding,” ACM Transactions on Knowledge Discovery from Data, vol. 14, no. 3, p. 27, 2020.
[6] H. Abdelmoum`ene, C. E. Bencheriet, H. Belleili, I. Touati, and C. Zemouli, “Dynamic matching optimization in ridesharing system based on reinforcement learning,” IEEE Access, vol. 12, pp. 29 525–29 535, 2024.
[7] J. Zhang, D. Wen, and S. Zeng, “A discounted trade reduction mechanism for dynamic ridesharing pricing,” IEEE Transactions on Intelligent Transportation Systems, vol. 17, no. 6, pp. 1586–1595, 2016.
[8] C. Yan, J. Yan, and Y. Shen, “Pricing shared rides,” in Proceedings of the 25th ACM Conference on Economics and Computation (EC ’24), 2024.
[9] A. A. Makhdomi and I. A. Gillani, “Fair and efficient ridesharing: A dynamic programming-based relocation approach,” ACM Transactions on Intelligent Systems and Technology, vol. 15, no. 5, p. Article 105, 2024.
[10] L. Chen, H.-N. Dai, X. Yuan, J. Huang, Y. Wu, and J. Wu, “D-spac: Double-sided preference-aware carpooling of private cars for maximizing passenger utility,” IEEE Transactions on Intelligent Transportation Systems, vol. 25, no. 8, pp. 9810–9827, 2024.
[11] B. Liu, Y. Dai, X. Li, W. S. Lee, and P. S. Yu, “Building text classifiers using positive and unlabeled examples,” in Proceedings of the 2002 IEEE International Conference on Data Mining (ICDM 2002). IEEE, 2002, pp. 179–186.
[12] C. Elkan and K. Noto, “Learning classifiers from only positive and unlabeled data,” in Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, 2008, pp. 213–220.
[13] R. Kiryo, G. Niu, M. C. du Plessis, and M. Sugiyama, “Positive-unlabeled learning with non-negative risk estimator,” in Advances in Neural Information Processing Systems, vol. 30, 2017.
[14] Z. Zhu, L. Wang, P. Zhao, C. Du, W. Zhang, H. Dong, B. Qiao, Q. Lin, S. Rajmohan, and D. Zhang, “Robust positive-unlabeled learning via noise negative sample self-correction,” in Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’23), 2023.
[15] S. Yadav and M. van der Schaar, “Censored quantile regression neural networks for distribution-free survival analysis,” in Advances in Neural Information Processing Systems (NeurIPS 2022), 2022.
[16] Y. Cao, X. Shen, X. He, Z. Zhang, and Y. Zhang, “Deep censored learning of the winning price in the real time bidding,” in Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. ACM, 2021, pp. 3080–3090.
[17] Z. Zhu, Q. Lin, H. Zhang, C. Du, B. Qiao, H. Dong, S. Rajmohan, and D. Zhang, “Predicting winning price in real time bidding with censored data,” in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. ACM, 2022, pp. 3859–3869.
[18] H. Chen, H.-L. Chen, and H.-T. Lin, “Arbitrary distribution modeling with censorship in real-time bidding advertising,” in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’22), 2022, pp. 2491–2501.
[19] Y. Zhang, J. Yang, Y. Ma, and M. Wang, “Bcorle(?): An offline reinforcement learning and evaluation framework,” in Advances in Neural Information Processing Systems (NeurIPS 2021), 2021.
[20] S. Garg, Y. Wu, A. Smola, S. Balakrishnan, and Z. C. Lipton, “Mixture proportion estimation and pu learning: a modern approach,” in Proceedings of the 35th International Conference on Neural Information Processing Systems, ser. NIPS ’21. Red Hook, NY, USA: Curran Associates Inc., 2021.