feature-engineering

Transform raw structured datasets into model-ready features with validation.

19|3|Updated Feb 28, 2026

Search Tags:#machine learning #feature engineering #data preprocessing #model performance #categorical encoding #date feature extraction #text features

Ready to run

npx skills add https://github.com/qa-aman/claude-skills --skill feature-engineering-qa-aman

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: feature-engineering
Source: https://github.com/qa-aman/claude-skills/tree/main/skills/by-role/data-scientist/feature-engineering
Command: npx skills add https://github.com/qa-aman/claude-skills --skill feature-engineering-qa-aman

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps data scientists transform raw data into meaningful features that substantially improve ML model performance, reducing the need for excessive tuning or additional data.

Core Features & Use Cases

Audit raw columns to identify transformation opportunities across numeric, categorical, datetime, and text data.
Apply numeric transformations such as log scaling, standardization, binning, and outlier capping to stabilize model inputs.
Encode categorical variables with methods suited to cardinality (one-hot for low cardinality, ordinal/target/frequency for higher cardinality).
Extract date and text features (hour, day-of-week, recency; TF-IDF or simple embeddings for text).
Create interaction and aggregate features to capture high-signal relationships and group-level effects.
Validate features to prevent leakage, assess feature importance, and document the final feature set with rationale.

Quick Start

Provide your dataset and start the feature engineering workflow to generate model-ready features.

Frequently Asked Questions about feature-engineering

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I transform raw data into model-ready features for machine learning?▼

To transform raw data into model-ready features, you must audit numeric, categorical, datetime, and text columns, then apply targeted transformations like scaling and encoding to boost machine learning performance.

What is the best way to encode categorical variables with high cardinality?▼

For high cardinality categorical variables, apply ordinal, target, or frequency encoding rather than one-hot encoding to prevent dimensional explosion while retaining predictive signal for model training.

How do I extract date and text features from raw dataset columns?▼

Extract date features by deriving hour, day-of-week, and recency values from datetime columns, and generate text features using TF-IDF or simple embeddings to capture semantic meaning.

How do I prevent data leakage when creating interaction and aggregate features?▼

Prevent data leakage during feature creation by validating features against the target variable, assessing feature importance, and documenting the final feature set with clear transformation rationale.

Can I use this feature engineering workflow on structured datasets containing mixed data types?▼

Yes, this feature engineering workflow processes structured datasets containing mixed numeric, categorical, datetime, and text columns by applying tailored transformations to each specific data type.

What numeric transformations should I apply to stabilize model inputs?▼

Apply numeric transformations such as log scaling, standardization, binning, and outlier capping to stabilize model inputs and reduce the need for excessive hyperparameter tuning.