Many social science questions have a spatial dimension. Exploring spatial patterns can help researchers generate hypotheses about how geography shapes social and environmental processes in their studies. This workshop is for beginners to learn a practical workflow for exploring spatial patterns of socio-demographic and environmental data. Through a hands-on case study, participants will retrieve socio-demographic data through the U.S. Census API and integrate it with spatial datasets. They will then use interactive visualization to explore geographic patterns and apply spatial statistical methods to identify spatial clusters, where nearby areas have similar values, and spatial outliers, where locations differ from their neighbors. Participants will learn to conduct these analyses in Google Colab using Jupyter notebooks based on Python. The cloud-based environment allows them to run analyses and scale computation on large datasets without local software installation. Basic familiarity with Python is helpful but not required.

Many social, environmental, and health processes exhibit spatial patterns. Nearby places may be related, and relationships between variables may vary across locations. When these spatial patterns are ignored, key assumptions of traditional statistical methods can be violated, leading to biased or misleading results.

This workshop introduces concepts and methods in spatial statistics and GeoAI for analyzing such patterns and relationships. Participants will learn about spatial dependence, where nearby observations tend to be related, and spatial heterogeneity and non-stationarity, where relationships between variables vary across space. Through a hands-on case study, participants will use mapping and spatial statistical methods to identify these patterns.

The workshop will also introduce XGBoost as a machine-learning approach for modeling complex and nonlinear relationships and use SHAP (Shapley Additive Explanations) to interpret how different variables contribute to model predictions in the social, environmental, and health sciences.

Analyses will be conducted in Python using Google Colab. Some familiarity with quantitative data analysis and spatial data science is encouraged. Prior experience with machine learning or programming is not required.

This workshop introduces directed acyclic graphs (DAGs) as a practical tool for selecting variables to include in analyses. Participants will learn how DAGs can help identify potential confounding and support transparent analytic decisions. Using examples from epidemiological research, the workshop will demonstrate how to develop and interpret DAGs, apply them to covariate selection, and perform sensitivity analyses to evaluate the robustness of the covariate selection process. This workshop is intended for students, researchers, and public health professionals with a basic understanding of regression modeling. No prior experience with DAGs or causal inference methods is required.