What process creates new sets of features to aid the training model in performing its task?

• A. Feature engineering
• B. Feature selection
• C. Dimensionality reduction
• D. Data normalization

Answer: (A)

Feature engineering is the process of creating new features from raw data to improve model performance. By deriving new attributes—such as combining two measurements into a ratio, applying a log transformation to handle skew, binning continuous variables, creating interaction terms, or aggregating information over time or groups—you provide the model with signals that can better capture patterns in the data. This step is guided by domain knowledge and data understanding, aiming to reveal relationships the model might not discover from the original features alone.

In contrast, selecting features means choosing which of the existing attributes to use, not creating new ones. Dimensionality reduction transforms features into a smaller set of components to reduce complexity, which can produce new representations but its primary goal is compactness rather than directly crafting signals from domain knowledge. Data normalization scales features for consistency but does not generate new features.