2022/04/19 by David Issa Mattos, Mattos, David Issa, Yuchu Liu +1 · 1 citation
Computer Science · #Advanced Software Engineering Methodologies #FOS: Computer and information sciences #Software Engineering (cs.SE) #Software Engineering Research #Software Reliability and Analysis Research
paper · pdf · doi:10.48550/arxiv.2204.08743
openalex publication_date 2022/04/19 · openalex created_date 2022/10/09 · openalex updated_date 2026/07/28
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.