How to determine whether a task is classification, regression, or matching?

The determination of the task type — classification, regression, or matching — depends on the nature of the data and the prediction goal:

• Classification — a task where an object needs to be assigned to one of several discrete classes. For example, determining whether an email is spam or not, or identifying the animal species in an image. The output data are categories or labels.

• Regression — a task of predicting a continuous numerical value. For example, forecasting house prices based on features or predicting tomorrow's temperature. The output is a number.

• Matching — a task of finding correspondences or pairs between elements from different sets. For example, matching resumes with job vacancies, finding similar products, or collaborative filtering in recommendation systems.

To determine the task type, answer the following questions:

• What type of output data is expected? Categories, numbers, or pairs?
• Is the goal to classify objects, predict numerical values, or find matches?

Examples:

• If the task is to determine which class an object belongs to, it is classification.
• If the goal is to predict a price, it is regression.
• If the task is to match users with products, it is matching.