Which of the following is an example of feature engineering?

• A. Converting timestamps into hour-of-day or day-of-week features.
• B. Normalizing the labels.
• C. Training with more epochs.
• D. Using a single constant feature.

Answer: (D)

Feature engineering is about creating new, informative features from existing data to help a model learn patterns. Turning a timestamp into hour-of-day or day-of-week features is a classic example because it reveals temporal structure and periodic patterns that the model can leverage, often improving predictive power. Normalizing the labels is a preprocessing step on the target and doesn’t add information to the inputs. Training with more epochs changes how the model learns but doesn’t introduce new input features. Using a single constant feature provides no extra information at all and can be ignored by the model. So, creating time-based features best illustrates feature engineering because it adds meaningful, learnable signals to the data.