Audit tabular datasets for data quality issues before modeling.
npx skills add https://github.com/rocklambros/rcs --skill auditing-data-quality
Please help me install this Agent Skill. Skill: auditing-data-quality Source: https://github.com/rocklambros/rcs/tree/main/skills/workflow/auditing-data-quality Command: npx skills add https://github.com/rocklambros/rcs --skill auditing-data-quality
This Skill audits a new tabular dataset before modeling so you can catch missingness, bad ranges, mislabeled columns, duplicates, and conflicting facts before they contaminate analysis or training.
Use the auditing-data-quality skill to inspect the attached dataset and return a full audit with shape, per-column stats, semantic classes, range checks, outlier flags, cardinality alarms, and row-level integrity findings.
High-intent search queries and answers about installing and using this skill.
To audit tabular data before modeling, run a dataset audit to generate per-column null counts, summary statistics, semantic typing, and outlier flags. This process catches missingness, bad ranges, and duplicates before they contaminate analysis or training.
Semantic typing infers whether a column is an ID, categorical, continuous, or datetime to match downstream handling with the data. It detects dataset anomalies by flagging suspicious cardinality, range violations, and conflicting fact pairs during the audit.
Yes, you can use this dataset audit on newly received CSV, Parquet, or DataFrame inputs. It processes bounded datasets to produce a go or no-go verdict by checking schema-shape changes and row-level integrity before fitting any model.
To find duplicate rows and conflicting facts in a DataFrame, apply row-level integrity checks. This identifies exact duplicates, same-content-different-ID duplicates, and conflicting fact pairs to prevent bad data from contaminating your analysis.
The best way to check for sudden metric regressions is to run a pre-training audit on bounded datasets. This flags outliers, range violations, and cardinality alarms, providing a comprehensive shape and per-column statistics report.
m0atCreate out-of-core DataFrames to browse and aggregate huge datasets efficiently with lazy evaluation and versatile IO formats.
brahiamsinhoOrchestrate multi-agent ML pipelines across data engineering, model training, deployment, and monitoring.
labs21-devProfile datasets to reveal structure, data quality, and relationships.
Patasse97Audit CSV and Parquet datasets for quality and generate remediation plans.
