1.3 A Typical Machine Learning Workflow

What we covered in this video lecture

In this lecture, we covered the overall supervised learning workflow. In supervised learning, we work with labeled training data. The labels typically come in two flavors: (1) categories (e.g., flower species) and (2) continuous values (e.g., house prices).

Additional content

While we discussed labeled data in the form of category labels or continuous values, there is a third common category of labels: ordinal labels. Ordinal labels are similar to category labels, but they imply a sorting order. As an intuitive example, think of disease severity, where the labels could be “severe” > “moderate” > “mild” > “no disease”.

Curious to learn more about ordinal methods? Check out the following resources:

• Conditional Ordinal Regression for Neural Networks (CORN) With Examples in PyTorch — YouTube tutorial by Sebastian Raschka
• Wenzhi Cao, Vahid Mirjalili, and Sebastian Raschka (2020) Rank Consistent Ordinal Regression for Neural Networks with Application to Age Estimation

Watch Video 1 Mark complete and go to Unit 1.4 →