· intermediate
Machine Learning with scikit-learn
The library and the workflow. Estimators, the fit and predict contract, why a pipeline is not a convenience but a correctness tool, and how preprocessing leaks the answer if you do it in the wrong order.
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
- Use the estimator interface, and say what fit, transform and predict each do
- Split data before touching it, and explain what that protects
- Build a pipeline so preprocessing is learned on training data only
- Encode categories and scale numbers without leaking the test set
- Choose a starting model from the problem rather than from fashion
- Search hyperparameters without quietly overfitting to the validation set
Before you start
- · Comfortable with pandas
- · Machine Learning Basics, or equivalent familiarity with features, labels and overfitting