πŸ—“οΈ Week 5: Non-linear algorithms and ensemble methods

Theme: Supervised Learning

Author

Welcome to the fifth week of this course!

This week, we wrap up our exploration of supervised learning techniques with some non-linear (classification) algorithms such as decision trees, random forests and support vector machines. We’ll also review how a procedure called cross-validation (which is a resampling technique) can be used to more robustly evaluate our models and to set their hyperparameters.

A homework formative about supervised learning techniques, in particular model evaluation and comparison as well as resampling (solutions released at the beginning of week 7) will be released this week. It’s a good preparation for your first summative, also released this week and due on Week 08! Stay tuned.

πŸ‘©πŸ»β€πŸ« Lecture Material

πŸŽ₯ Looking for lecture recordings? You can only find those on Moodle, typically a day after the lecture. If you can’t find the recordings, please contact πŸ“§ .

Download the Quarto markdown file for this lecture:

Important

Make sure to switch to the Python 12 environment from this week onwards (see this page for details)

You might be missing a few libraries for this lecture. You can install them by running

conda install -c conda-forge imbalanced-learn shap missingno