What is binary classification?

This article explains the Machine Learning term Binary Classification and Binary Classification as a concept within the AI & Analytics Engine

Binary Classification is a type of Classification task, where there are only two possible classes or outcomes to predict. Additionally, between the two possible classes, one class is the outcome of interest. The outcome of interest is known as a positive class label. In simpler terms, it is a kind of problem where one needs to accurately predict the answer to a yes or no question about entities.

Some examples of Binary Classification are:

• Predicting whether a customer will default his mortgage repayments,

• Predicting whether a person will exit his internet plan upon expiry of his contract, and

• Predicting whether a credit-card transaction is normal or the result of card theft

Binary Classification within the AI & Analytic Engine

Within the Engine, while creating an app, if the user chooses the “Predict a variable/column” problem type, Binary Classification is available as the “Answer a yes/no question” problem sub-type.

This option is automatically recommended by the Engine when there are only two possible outcomes. The positive class label is automatically chosen to be the rarer of the two class labels, based on their frequencies of occurrence in the dataset.