Feature engineering
Transforming raw data into inputs a model can learn from more effectively, such as encoding a category, scaling a numeric range, or deriving a new column from existing ones. It is done before training and evaluated by whether it improves validation performance, not by whether it looks reasonable.
Why exams ask this
Tested as a domain-knowledge question disguised as a technical one. The correct answer usually involves deriving a feature from what is known about the problem, such as a ratio or a time-based aggregate, rather than feeding every raw column into the model unchanged.
Relevant to
Related concepts
Resources
No resources linked to this concept yet.