Unsupervised Learning

Estimated time: 5 hours

Read Chapter 9: Unsupervised Learning Techniques from Hands-On Machine Learning with Scikit-Learn, Keras & TensorFlow, up to but not including Gaussian Mixture Models.

The focus is clustering and discovering structure in data without labelled target values.

After completing the reading, you should be able to explain the main difference between supervised and unsupervised learning and recognize important clustering approaches such as K-Means and DBSCAN/HDBSCAN.

Use the three-pass method below.

First Pass: Get the Big Picture — 45 minutes

Orient yourself in the chapter.

• Read the section titles and subheadings up to Gaussian Mixtures.
• Look at the figures, plots, and diagrams and read their captions.
• Skim the conceptual exercise questions at the end of the chapter.

Goal: Build a mental map of unsupervised learning and how algorithms can discover structure in unlabelled data.

Second Pass: Read for Understanding — 3 hours

Read the relevant sections in detail.

Stop before Gaussian Mixtures.

If you encounter algorithms or formulas that you do not fully understand, keep moving. The goal is to understand what each method is for and how it approaches the problem rather than mastering all of the mathematics.

Goal: Understand how unsupervised-learning methods identify structure and patterns in data.

Third Pass: Conceptual Exercises — 1 hour

Return to the conceptual questions at the end of the chapter.

For each question:

• think through your answer;
• locate the relevant part of the chapter;
• check whether you can explain the concept in your own words.

Focus on the conceptual questions rather than the coding exercises.

Goal: Strengthen your understanding by connecting questions back to the relevant concepts in the chapter.