« Back to Glossary Index
Feature Engineering /ˈfiː.tʃər ˌɛn.dʒɪˈnɪə.rɪŋ/ noun — In machine learning and data science, feature engineering is the process of creating, transforming, selecting, or extracting input features from raw data to enhance a model’s predictive performance. It is a critical step in the data preparation pipeline, often having a greater impact on model accuracy than the choice of algorithm itself.
This process involves:
- Cleaning and formatting raw data
- Combining features to reveal hidden patterns (e.g. date → day, month, year)
- Encoding categorical variables (e.g. one-hot encoding)
- Scaling or normalizing numerical features
- Deriving new variables that better represent the underlying problem
Feature engineering requires a blend of domain knowledge, mathematical insight, and statistical understanding. It directly affects how well a machine learning model can generalize to unseen data, making it essential for robust and interpretable AI systems.
« Back to dictionary
