cleaning-data

To clean messy data, this skill automates duplicate detection, outlier remediation, NULL handling, and free-text categorization using SQL-based transformations, producing analysis-ready datasets with audit-ready logging.

The best way to detect outliers and standardize free-text categories is through the skill's phase-guided workflow, which systematically scopes, detects, and executes SQL transformations while tracking the rationale for each remediation step.

You handle NULL values and validate data quality by running the cleaning pipeline within DataPeeker sessions, which applies SQL-based validation and remediation strategies to generate clean tables and quality reports.

Yes, you can use SQL-based transformations for deduplication without external dependencies, as the skill operates independently to detect and remediate duplicates while maintaining an audit trail for reproducibility.

Yes, the data cleaning pipeline supports reproducible workflows by employing a 5-phase process—scope, detection, strategy, execution, and verification—that logs rationale and maintains an audit trail for all transformations.

You should run data cleaning before guided-investigation or exploratory-analysis workflows to produce clean_[table] outputs, ensuring downstream analyses operate on validated, standardized, and analysis-ready datasets.