Can I define custom ratios for my data subsets?

Yes, you can specify exact percentages for training, validation, and testing sets, such as 80/10/10 or 70/30 splits.

Does this skill prevent data leakage?

By creating distinct, non-overlapping files for training and testing, the skill helps enforce proper data hygiene to prevent leakage during the evaluation phase.

Do I need to manually write the splitting logic?

No, you simply describe your desired split in natural language, and the skill generates and executes the necessary Python code for you.

What file formats does the Dataset Splitter support?

The skill generates Python code typically designed to process structured data formats like CSV, which are standard in machine learning workflows.

How does this skill ensure data randomization?

The generated code utilizes industry-standard libraries to shuffle and randomize data during the split to ensure that subsets are representative and unbiased.

Machine Learning Dataset Splitter

Name: Machine Learning Dataset Splitter
Author: jeremylongshore

byjeremylongshore

•

883

データサイエンスとML

Automates the partitioning of datasets into training, validation, and testing sets for machine learning workflows.

概要

This skill streamlines data preparation by automatically dividing datasets into the specific subsets required for robust machine learning model development. By analyzing user-defined ratios and generating Python-based splitting logic using standard libraries, it ensures data integrity, randomization, and optional stratification, making it an essential tool for data scientists and developers aiming to evaluate model performance accurately within the Claude Code environment.

主な機能

883 GitHub stars
Automated partitioning into training, validation, and testing sets
Customizable split ratios (e.g., 70/15/15 or 80/20 distributions)
Maintains data integrity across CSV and other structured formats
Generation of executable Python code using standard ML libraries
Randomization logic to prevent selection bias in subsets

ユースケース

Partitioning large data files for final model performance benchmarking
Preparing raw CSV datasets for initial machine learning model training
Creating dedicated validation sets for hyperparameter tuning

概要

主な機能

883 GitHub stars
Automated partitioning into training, validation, and testing sets
Customizable split ratios (e.g., 70/15/15 or 80/20 distributions)
Maintains data integrity across CSV and other structured formats
Generation of executable Python code using standard ML libraries
Randomization logic to prevent selection bias in subsets

ユースケース

Partitioning large data files for final model performance benchmarking
Preparing raw CSV datasets for initial machine learning model training
Creating dedicated validation sets for hyperparameter tuning