Despite their current growth and future promise, ride-hailing companies struggle with brand loyalty. As a result, they spend heavily on various marketing tools, especially promotional offers, to encourage passenger engagement, which is often measured by how frequently passengers ride through their platforms. Although extensive research has investigated the passenger intention to continue using ride-hailing services, research that explicitly models the dynamics of passenger engagement is very scarce. In this research, we propose to capture passenger engagement dynamics in ride-hailing services and the factors contributing to them. We combine a heterogeneous hidden Markov model framework with Poisson regression models to probabilistically analyze the transition processes of passenger engagement. Specifically, we capture the influences of various promotional offers on the engagement transition probabilities. We conduct numerical experiments using real-world ride-hailing data. Results show that our model identifies inactive, occasional, and active engagement levels. Our coefficient estimates and sensitivity analysis show that giving moderately more promotional offers to inactively and occasionally engaged passengers would efficiently activate them. More importantly, we derive information about which promotional offers have more significant impacts on the passengers of different engagement levels.

尽管他们目前的增长和未来的前景，乘车公司在品牌忠诚度方面苦苦挣扎。因此，他们在各种营销工具上投入巨资，尤其是促销优惠，以鼓励乘客参与，这通常以乘客使用其平台的频率来衡量。尽管广泛的研究调查了乘客继续使用叫车服务的意愿，但明确模拟乘客参与动态的研究非常少。在这项研究中，我们建议捕捉乘车服务中的乘客参与动态及其影响因素。我们将异构隐马尔可夫模型框架与泊松回归模型相结合，以概率分析乘客参与的转变过程。具体来说，我们捕获了各种促销优惠对参与转换概率的影响。我们使用真实世界的叫车数据进行数值实验。结果表明，我们的模型可以识别不活跃、偶尔和活跃的参与水平。我们的系数估计和敏感性分析表明，向不活跃和偶尔参与的乘客提供适度更多的促销优惠会有效地激活他们。更重要的是，我们获得了关于哪些促销优惠对不同参与度的乘客有更显着影响的信息。我们的系数估计和敏感性分析表明，向不活跃和偶尔参与的乘客提供适度更多的促销优惠会有效地激活他们。更重要的是，我们获得了关于哪些促销优惠对不同参与度的乘客有更显着影响的信息。我们的系数估计和敏感性分析表明，向不活跃和偶尔参与的乘客提供适度更多的促销优惠会有效地激活他们。更重要的是，我们获得了关于哪些促销优惠对不同参与度的乘客有更显着影响的信息。