What happens if the review finds issues?

The skill provides a structured report categorizing issues by confidence level and includes specific code snippets to fix the identified problems.

Does this skill check for code style or linting issues?

No, the skill specifically ignores coding style and personal preferences to focus exclusively on statistical validity, data integrity, and reproducibility.

What is the Iron Law of DS Review?

It is a strict requirement that the skill only reports methodology or data issues where it has 80% or higher confidence, preventing false positives and subjective style critiques.

Can it detect data leakage in machine learning models?

Yes, it is designed to identify common data science errors including data leakage, selection bias, p-hacking, and training-serving skew.

Data Science Review Specialist

Name: Data Science Review Specialist
Author: edwinhu

byedwinhu

0•

Ciencia de Datos y ML

Validates data analysis workflows by reviewing methodology, data quality, and statistical integrity with a high-confidence scoring system.

The ds-review skill is a specialized auditing tool designed for the validation phase of data science workflows. It ensures that analyses are not only mathematically correct but also reproducible and statistically sound. By enforcing the 'Iron Law of DS Review,' it filters out subjective style preferences and minor nitpicks, focusing instead on critical issues like data leakage, selection bias, and improper handling of missing values. It provides structured feedback, confidence scoring—reporting only issues with 80% or higher certainty—and actionable fixes to ensure data-driven decisions are based on valid foundations.

Características Principales

01Statistical methodology verification and assumption checking

02Reproducibility assessment for seeds, versions, and transformations

03Comprehensive data quality audits for missing values, outliers, and types

04Structured markdown review reports with actionable code fixes

05Confidence-based issue reporting with a strict 80% threshold

060 GitHub stars

Casos de Uso

01Conducting a final quality assurance check on exploratory data analysis before stakeholder delivery

02Ensuring complex statistical notebooks are fully reproducible with documented seeds and dependencies

03Validating a machine learning model's training pipeline for leakage or bias

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add edwinhu/workflows ds-review

For use in Claude.ai and ChatGPT

Download Skill