Compare two dataset snapshots and fail your workflow if the change is
statistically significant — a drop-in CI/CD wrapper around the same
root-cause + Welch's t-test / chi-square engine that powers "What Changed"
in the full app. Pure pandas/scipy: no AI key required unless you opt into
a narrated summary.
Usage
- name: Check for significant data drift
uses: NaiaLorente/data-analyst@v1.0.0
with:
file-a: data/last_week.csv
file-b: data/this_week.csv
metric: revenue
segment: region
fail-on-significant: "true"