Causal modeling of energy and resource demand
摘要
Many sustainability issues that our society is facing stem from an excessive demand for energy and resources (E&R). Modeling E&R demand is important to understand how people consume E&R and to tailor policies to curb E&R demand. Traditionally, statistical models and machine learning have been used to model E&R demand. While these models can perform well, they are based on associations (i.e., correlations) and cannot tell which variable causes another. The Science of Causality has emerged as a promising field to build causal models and capture causal relationships. Causal models offer a graphical representation of the flow of causality and a quantitative estimate of the causal effects between the variables. This article offers an introduction to the Science of Causality within the context of E&R demand. It applies causal discovery and inference methods to three types of E&R: transport (travel mode choice), electricity use, and water consumption. It ends with a discussion of opportunities and limitations of current causal methods to model E&R demand. Overall, causal discovery and inference offer much potential to detect the underlying relationships that drive E&R demand and identify variables that should/should not be intervened upon to lower E&R demand for a more sustainable future.