Skip to main content

Feature Engineering

Updated May 12, 2023 ·

Feature Engineering​

After designing a machine learning model, the next step is feature engineering.

Feature engineering involves selecting, modifying, and creating useful variables (features) from raw data. These features help the model make better predictions.

  • Feature: A measurable property, like a column in a table.
  • Raw vs. Engineered Features: We can use existing data or create new features.

For more information, please see Feature Engineering.

Feature Store​

A feature store is a central location to save and reuse features across projects. It helps large teams maintain consistency, but smaller projects may not need it.

For more information, please see Feature Stores.

Data Version Control​

Just like Git tracks code changes, data version control tracks dataset changes.

  • Keeps history: Allows rolling back to previous versions.
  • Ensures consistency: Prevents unexpected changes in model inputs.

Feedback