analyze-data-quality

To check data quality before modeling, assess datasets for completeness, uniqueness, schema drift, and data leakage. This Skill evaluates ML feature datasets to identify null rates, distribution shifts, and broken joins, delivering structured findings with severity ratings and remediation steps to prevent flawed analysis.

Validating data integrity for dashboards involves checking fact and dimension tables for stale data, missing values, and referential integrity. This Skill assesses query results and dataframes for grain mismatches and source mismatches, providing risk-aligned reports with likely root causes to ensure dashboard reliability.

Detecting schema drift and distribution shifts requires evaluating event datasets for structural changes and statistical anomalies. This Skill profiles data pipelines to identify schema drift, backfills, and grain mismatches, returning actionable automated test recommendations to reduce analytical risk.

Yes, you can assess referential integrity and duplicates across tables, files, and dataframes. This Skill performs comprehensive quality checks for validity and consistency across event, dimension, and experiment datasets, highlighting broken joins and assigning severity ratings to guide remediation.

Broken analysis often happens due to untrustworthy datasets with hidden grain mismatches, null rates, or data leakage. You can prevent it by using data profiling to assess dataset trustworthiness, identifying likely root causes for quality issues, and implementing recommended automated tests before critical use.

You should perform dataset validation before using data for downstream analysis, modeling, or pipeline processing. This Skill eliminates guesswork by running risk-focused quality checks on files and dataframes, delivering structured findings with severity ratings to mitigate analytical risk prior to execution.