This paper explores new methods for regression models to understand the evolution of headway irregularity of public transport services. In particular, this article is structured around an urban bus service context. This is relevant as headway variability has an important effect on travel time reliability. To do so, a Dynamic Panel Model was estimated to account for serial correlation, which is a novel approach in this context. The coefficient of variation of headways was selected as the dependent variable due to its direct relationship with excess waiting time. The results indicate that the most significant source of headway irregularity is the dispatch at the terminals. The irregularity of uneven dispatches propagates and worsens downstream. To prevent headway variability increasing, the most effective measures are segregated corridors and off-board payment stops. This should be useful to orient interventions in the system’s operations, infrastructure, and contracts that will improve reliability the most.

本文探索了用于回归模型的新方法，以了解公共交通服务的车速不规律性的演变。特别是，本文围绕城市公交服务环境进行了结构设计。这一点很重要，因为车距变化会对行驶时间的可靠性产生重要影响。为此，估计使用动态面板模型来说明序列相关性，这在此情况下是一种新颖的方法。由于车头的变化系数与等待时间的长短直接相关，因此选择车头的变化系数作为因变量。结果表明，最大的车距不规则来源是码头的调度。调度不均的不规则性在下游传播并恶化。为了防止车头时速变化性增加，最有效的措施是隔离走廊和车外支付站。这将有助于对系统的操作，基础结构和合同进行干预，从而最大程度地提高可靠性。