Phase 3 · Machine learning · Foundations
Preparing data
Clean, encode and scale data before a model sees it.
Open this lesson Create a free account
Why it matters
Models only accept complete, numeric input, and real data is neither: customers who never subscribed have no email open rate, regions are words, and order values are in dollars while discount share is a fraction. Data preparation usually takes more time than modeling, and small mistakes here (like computing an average over all your data, test set included) quietly inflate results.
What you'll be able to do
- Find and handle missing values: drop, fill, and flag
- Turn categories into numbers with one-hot encoding
- Scale numeric features with standardization
- Fit every preparation step on training data only, and reuse it
Before this lesson
Next in this phase: Feature engineering