data-analysis

Data Analysis Workflow

Run an end-to-end data analysis in R or Python: load, explore, analyze, and produce publication-ready output.

Input: $ARGUMENTS — a dataset path (e.g., data/county_panel.csv) or a description of the analysis goal (e.g., "regress wages on education with state fixed effects using CPS data").

Before making analysis changes, read ${CLAUDE_PLUGIN_ROOT}/rules/orchestrator-protocol.md, ${CLAUDE_PLUGIN_ROOT}/rules/analysis-verification.md, and ${CLAUDE_PLUGIN_ROOT}/rules/quality-gates.md; for replication or extension requests, also read ${CLAUDE_PLUGIN_ROOT}/rules/replication-protocol.md.

Phase 0: Choose Language

R is the default. If ambiguous, use AskUserQuestion to pick R (Recommended — tidyverse/fixest/ggplot2, full plugin support including r-reviewer) or Python (pandas/statsmodels). Pick R when the project has existing .R code; Python when it has existing .py or .ipynb.

Phase 0.5: Ingest Non-Tabular Inputs

If $ARGUMENTS points to a PDF, HTML page, government portal URL, or other non-tabular source, Read ${CLAUDE_PLUGIN_ROOT}/skills/data-analysis/references/ingest-recipes.md and follow it before Phase 1. Skip this phase for flat files (.csv, .parquet, .dta, .rds, .feather, .json).

R Track

Constraints

• Follow ${CLAUDE_PLUGIN_ROOT}/rules/r-code-conventions.md for all standards
• Save scripts to scripts/R/ with descriptive names
• Save all outputs (figures, tables, RDS) to output/
• Use saveRDS() for every computed object
• Run r-reviewer on the generated script before presenting results

Phase 1: Setup and Data Loading

1. Create R script with proper header (title, author, purpose, inputs, outputs) — Read ${CLAUDE_PLUGIN_ROOT}/skills/data-analysis/assets/analysis-script.R and adapt
2. Load required packages at top (library(), never require())
3. Set seed once at top: set.seed(42)
4. Create output directories: dir.create("output/analysis", recursive = TRUE, showWarnings = FALSE)
5. Load and inspect the dataset