Machine Learning for Industrial Applications

8Binary and Beyond Binary Classification

Binary and beyond binary classifications are mainly classified as follows:

1. Classification
2. Scoring and ranking
3. Class probability estimation
4. Handling more than two classes
5. Descriptive learning

Descriptive learning theories make statements about how learning occurs and devise models that can be used to explain and predict learning results [1]. When describing different descriptive theories of learning below, we will follow the common categorization that distinguishes between behaviorist, cognitive, and constructive learning theories.

8.1 Binary Classification

A typical machine learning challenge is classifying data into one of two categories. You might need to foretell if a consumer is likely to make a purchase, whether a credit card transaction was fraudulent, whether deep space signals indicate the existence of a new planet, or whether a medical test indicates the presence of a disease. These are all difficulties of binary categorization.

Binary classification tasks with two class labels are referred to as binary classification.

Examples comprise the following:

• Detection of spam email (spam or not).
• Churn forecast (churn or not).
• Prediction of conversion (buy or not).

Binary classification problems often require two classes—one representing the normal state and the other representing the aberrant state.

For instance, the normal condition is “not spam,” whereas the abnormal state is “spam.” Another illustration is when a task involving ...