Methodology of our research

Our CHES methodology features both data science informed and disciplinary theory-driven approaches. From the data science perspective, we leverage both traditional (e.g., regression ananlysis, spatial statistics) and non-traditional (e.g., a digital, high performance, and 4-D holographic framework, agent-based modeling, convolutional neural networks, graph neural networks, landscape survival analysis) for systems representation and modeling. From the theory-driven perspective, we follow not only traditional inductive, deductive, and abductive (mostly used in social sciences) reasoning, but also a complex adaptive systems approach--a “new kind of science” (Wolfram 2002) paradigm. In the context of complex adaptive systems, our research highlights high dimensionality, hierarchical structure, heterogeneity, nonlinear relationships, feedback, path dependence, emergence, equifinanlity, and multifinality, with an aim of harnessing (rather than ignoring or eliminating) system complexity such that innovative actions might be taken to steer the CHES under investigation in beneficial directions (An et al. in review). Below is a list of exemplar methods we often rely on.

As new methods, technologies, or projects emerge, our methodology will surely evolve and become more sophisticated. We keenly welcome more researchers and practitioners to join our endeavors. We believe that methodological explorations are not only useful, but also fun!