Feature Engineering
The process of selecting, transforming, and creating input variables (features) from raw data to improve model performance. It requires domain knowledge to identify what information is most useful for the model.
Why It Matters
Feature engineering can be the difference between a mediocre and an excellent model. Even simple feature transformations often outperform more complex model architectures.
Example
Converting a raw timestamp into features like hour-of-day, day-of-week, and is-holiday for a demand forecasting model — these derived features capture patterns the raw timestamp cannot.
Think of it like...
Like a chef preparing ingredients before cooking — chopping, marinating, and combining raw ingredients into the right form makes the final dish much better.
Related Terms
Dimensionality Reduction
Techniques that reduce the number of features (dimensions) in a dataset while preserving the most important information. This makes data easier to visualize, speeds up training, and can improve model performance.
Data Preprocessing
The process of cleaning, transforming, and organizing raw data into a format suitable for machine learning. This includes handling missing values, encoding categories, scaling features, and removing outliers.