Exploratory R analysis for raw local datasets before a formal design is fixed. Use when data, measurement, variable roles, hypotheses, or methods are uncertain and Codex should inspect/clean data, infer candidate roles, map statistical/econometric/network/time-series/QCA/interpretable-ML models, consider DAGs for causal-control questions, produce notebooks under notebooks/exploratory, R/Quarto/QMD files, 300 dpi TIF figures, Chinese Markdown reports, and Obsidian backups. Project reuse is same-skill only: same data used with exploratory-r-analysis and ssci-r-analysis-writing are separate projects.
