论文标题
关于使用因果图形模型在汽车域中设计实验
On the Use of Causal Graphical Models for Designing Experiments in the Automotive Domain
论文作者
论文摘要
随机现场实验是评估软件变化对客户的影响的黄金标准。在在线域中，随机化一直是确保交换性的主要工具。但是，由于部署条件的不同和对周围环境的高依赖性，对汽车软件的设计实验需要考虑更高数量的限制变量，以确保有条件的交换性。在本文中，我们展示了如何利用因果图形模型设计实验并明确传达实验假设的沃尔沃汽车。这些图形模型用于进一步评估实验有效性，计算直接和间接因果效应，以及对因果结论的可运输能力的理由。
Randomized field experiments are the gold standard for evaluating the impact of software changes on customers. In the online domain, randomization has been the main tool to ensure exchangeability. However, due to the different deployment conditions and the high dependence on the surrounding environment, designing experiments for automotive software needs to consider a higher number of restricted variables to ensure conditional exchangeability. In this paper, we show how at Volvo Cars we utilize causal graphical models to design experiments and explicitly communicate the assumptions of experiments. These graphical models are used to further assess the experiment validity, compute direct and indirect causal effects, and reason on the transportability of the causal conclusions.
