how-to-data-qualitylisted

Put a data-quality gate on a warehouse load - map every defect to one of the six DQ dimensions (completeness, validity, consistency, timeliness, uniqueness, accuracy), run a real FAIL -> fix -> PASS cycle with a quarantine ledger that reconciles every row, and generate a self-contained executive DQ scorecard where every number is computed. Use when asked to "check data quality", "validate this load", "build DQ checks", "data quality scorecard", "should we trust this data", "set up a quality gate", "quarantine bad rows", or when a pipeline needs defects caught at ingestion instead of in a board meeting. Walks the 6-step pipeline - input, sample data, objective, find-skills, build (gate + fixes + scorecard), expert review.
phoebefu6/phoebe-data-skills · ★ 1 · Data & Documents · score 72

Install: claude install-skill phoebefu6/phoebe-data-skills

# how-to-data-quality Data-Analytics-layer skill (layer 2 of the phoebe-data-skills 4-layer roadmap). The job is not "run some checks" - it is a **gate with a verdict**: does the load ship, or is it blocked, and where did every row go? Showcase walkthrough (Everrest retail case, real FAIL -> PASS run): https://github.com/phoebefu6/phoebe-data-skills - `docs/how-to-data-quality/` ## Where this sits in the lineage **raw dump -> [THIS GATE] -> warehouse** -> marts -> scorecard/agent. Quality problems are ingestion problems; catch them where they enter, with a ledger, not downstream where they surface as a wrong board number. ## The three rules that separate it from every DQ tutorial 1. **A gate, not a report.** Any BLOCKER failure blocks the load, whatever the average score says. A DQ score that averages its way past a blocker is decoration. 2. **Every row is accounted for.** rows_in = rows_loaded + rows_quarantined (+ merged). Quarantine with reasons; never silently drop. 3. **Route fabrication, never repair it.** A suspicious amount is evidence for an investigation, not a formatting defect. Hold it out of finance rollups and preserve it untouched. A pipeline that "corrects" fabricated data is lying twice. ## The six dimensions, as checks | Dimension | Check pattern | Everrest example | |---|---|---| | completeness | required attributes present | orphan customer refs surface as nulls | | validity | accepted values / types / ranges | 14 labels arrive for