Olive R. Cawiding, Kangmin Lee
School of Transdisciplinary Studies, KAIST, Daejeon, Republic of Korea
ABSTRACT
Time series data are ubiquitous in science and public health, capturing the temporal evolution of complex systems and containing rich information about trends, interactions, and underlying mechanisms. Extracting this information requires appropriate preprocessing, and detrending and denoising play a crucial role in separating meaningful dynamical signals from long-term trends and noisy fluctuations that can obscure true interactions; these concepts will be introduced in this workshop. This workshop further presents GOBI, a model-based causal inference method designed to infer directional relationships between variables from time series data. GOBI overcomes key limitations of conventional causal inference approaches, particularly those arising from synchrony and indirect effects. This method has been successfully applied to uncover causal links between weather variables and dengue incidence in the Philippines. In this hands-on workshop, participants will learn to apply detrending techniques and GOBI to real-world data to infer relationships between weather factors and dengue fever incidence in Singapore.