What is the primary purpose of feature engineering in machine learning?

Correct answer: To transform raw data into features that better represent the underlying patterns

Option A is correct because feature engineering is the process of using domain knowledge to transform or construct input variables from raw data so that they more effectively capture the underlying patterns the model needs to learn, directly improving predictive performance. Option B describes hyperparameter tuning, which adjusts model configuration settings such as learning rate or tree depth and is a separate optimization step. Option C describes algorithm selection, which is the process of evaluating different model architectures for a given problem, not the transformation of input data. Option D describes train-test splitting, which is a data partitioning technique used to evaluate generalization and is distinct from feature construction.