Definition
Feature Engineering is the process of creating, selecting and transforming features used as inputs by Machine Learning models to improve their usefulness, performance and predictive quality.
Human Explanation
It prepares data so that a model can more effectively recognise relevant relationships and patterns.
Why it Matters
The relevance and quality of features often affect model performance more than the choice between similar algorithms.
Conceptual Boundary
Feature Engineering changes the representation of model inputs, not the underlying learning algorithm. It may include Feature Selection, but selecting existing features and creating new ones remain distinct activities within the broader process.
Practical Perspective
The process may include selecting variables, deriving new features, transforming values, encoding categories and scaling numerical data. The same transformations must be applied consistently during training and prediction, using only information available at the time of prediction to prevent data leakage.