· intermediate
Feature Engineering
Most of the improvement in a model comes from what you feed it. Making features from dates, text and history, the ones that quietly contain the answer, and knowing which to keep.
Written in house, from the work rather than from a syllabus. Every example is one we would use ourselves, and every quiz question explains why its answer is right.
What you’ll learn
- Turn a raw column into features a model can use
- Extract signal from dates, including cyclical encoding
- Aggregate history into per-entity features without crossing the time boundary
- Recognise target leakage, and test for it when a score looks too good
- Handle high-cardinality categories without exploding the feature count
- Read feature importance without over-trusting it
Before you start
- · Comfortable with pandas
- · Machine Learning Basics, or equivalent