3.4. Reflection Checkpoint: Machine Learning Mastery

Skipping this checkpoint risks carrying gaps forward into more complex topics. Without solid ML fundamentals, you'll struggle when Computer Vision, NLP, and Generative AI questions reference these underlying techniques. Imagine reaching an exam question about "predicting categories" and not being certain whether that's regression or classification.

Consider these questions like a self-diagnostic. For instance, the logistic regression question trips up many people because the name is misleading. If you hesitate, review Section 3.1. These checkpoints exist because mastery means instant recognition, not working through logic each time.

1. A retailer wants to group online shoppers by similar attributes for targeted marketing. Which ML technique?

   • Clustering. There are no predefined categories (labels)—we're finding natural groups in the data.

2. You need to predict the probability of diabetes based on age and body fat percentage. What should the model include?

   • Two features (age, body fat percentage) and one label (probability of diabetes).

3. What is logistic regression used for?

   • Classification! Despite the name, it predicts categories (yes/no), not continuous numbers.

4. Which Azure ML capability automatically selects the best algorithm?

   • Automated Machine Learning (AutoML).

5. Why split data into training and validation sets?

   • To test whether the model can generalize to new, unseen data—not just memorize training examples.