quant-explorationlisted

# quant-exploration — Exploratory Data Analysis ## Objective Give the user a thorough visual and numerical overview of their cleaned dataset before any inferential analysis. The output should make distributions, outliers, missing patterns, and relationships between variables immediately visible — for both technical analysts and non-technical stakeholders. --- ## Pre-flight **Required inputs:** - `output/data/cleaned_data.csv` — cleaned dataset from `quant-data-prep` - `output/data/column_metadata.json` — column types, labels, weight_col **If these files don't exist:** tell the user to run `/quant-data-prep` first. **Fallback:** if the user provides a raw data path without running data-prep, offer to run a minimal audit (data-prep Step 1 + Step 2 structural checks only) before proceeding. --- ## Pipeline ### Step 1 — Load data and metadata ```python # See references/eda_plots.py → load_prepared_data() ``` - Load `cleaned_data.csv` via pandas - Load `column_metadata.json` for types, labels, weight_col - Classify columns: ordinal/continuous → numeric summary; nominal/binary → frequency table - Report: n rows, n columns, weight status ("weighted" / "unweighted") --- ### Step 2 — AskUserQuestion: grouping variable (optional) ``` AskUserQuestion: question: "Would you like to split all visualizations by a grouping variable (e.g., gender, brand, market)?" header: "Grouping" options: - label: "No grouping — show overall distributions (Recommended f