Does this skill help with model reproducibility?

Yes, it specifically checks for the presence of set.seed() before random operations like splitting and resampling to ensure results can be consistently reproduced.

How does this skill help with imbalanced datasets?

It identifies missing stratification arguments in initial_split() and vfold_cv(), and flags the use of simple accuracy when metrics like ROC AUC or F1-score are more appropriate for the data distribution.

Why should I use workflows instead of manual prep and bake?

Workflows bundle preprocessing and modeling into a single object, automatically ensuring that training statistics are applied correctly to new data, which minimizes the risk of human error and data leakage.

What common data leakage errors does this skill detect?

It detects patterns such as calling prep() on test data, performing feature selection before data splitting, and applying normalization or scaling to the entire dataset before initial_split().

Tidymodels Code Review Patterns

Name: Tidymodels Code Review Patterns
Author: choxos

bychoxos

0•

データサイエンスとML

Analyzes R machine learning code to detect data leakage, resampling violations, and workflow anti-patterns using tidymodels principles.

This skill provides systematic code review for R projects using the tidymodels ecosystem, specifically targeting critical errors like data leakage and resampling violations. Based on 'Tidy Modeling with R' (TMwR) principles, it helps developers identify anti-patterns such as preprocessing before data splits, fitting recipes on test data, and missing stratified sampling. By enforcing standardized workflow patterns and robust evaluation metrics, it ensures machine learning models are reproducible, statistically sound, and correctly validated for production environments.

主な機能

01Validates evaluation metrics to ensure appropriateness for imbalanced datasets.

02Identifies improper resampling techniques and missing stratified sampling.

03Detects critical data leakage patterns in R recipes and workflows.

04Ensures reproducibility by checking for consistent random seed usage.

05Flags manual preprocessing anti-patterns to promote workflow-based modeling.

060 GitHub stars

ユースケース

01Refactoring manual R preprocessing and modeling code into standardized tidymodels workflows.

02Reviewing R machine learning scripts to ensure scientific validity and prevent overoptimistic results.

03Auditing biostatistical models for data leakage and proper cross-validation setups.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add choxos/biostatagent tidymodels-review-patterns

For use in Claude.ai and ChatGPT

主な機能

01Validates evaluation metrics to ensure appropriateness for imbalanced datasets.

02Identifies improper resampling techniques and missing stratified sampling.

03Detects critical data leakage patterns in R recipes and workflows.

04Ensures reproducibility by checking for consistent random seed usage.

05Flags manual preprocessing anti-patterns to promote workflow-based modeling.

060 GitHub stars

ユースケース

01Refactoring manual R preprocessing and modeling code into standardized tidymodels workflows.

02Reviewing R machine learning scripts to ensure scientific validity and prevent overoptimistic results.

03Auditing biostatistical models for data leakage and proper cross-validation setups.