Machine Learning · Lesson 8
Feature Engineering for Machine Learning
Learn to prepare raw business data for machine learning with missing-value handling, categorical encoding, scaling, derived features, and leakage prevention.
What is feature engineering?
Turn raw columns into information a model can use.
A feature is an input used by a machine learning model. Feature engineering means selecting, transforming, or creating input variables so they represent useful patterns in the problem. Good features can make a model easier to train and more useful; extra columns are not automatically better.
Business meaning
For a delivery-time model, distance, traffic period, and order size may be more useful than an arbitrary order reference number.
Delivery time rises steadily as distance grows: a clear pattern.
Known when the estimate is made? Yes
Illustrative delivery data. Select a field to plot its values against delivery time, and check whether it is known when the estimate is made.
Try it
A delivery company wants to estimate delivery time. Inspect the field explorer above, then decide: which field is most likely to be a useful predictive feature?
Quick knowledge check
What is the main purpose of feature engineering?