Basic machine learning terminology

One of the most confusing to beginners things in ML is its terminology. Every ML concept has several synonyms. What’s makes things even worse, each ML course or book uses different combination of these. There’s no common standard.

Below, there’s a list of terms and their possible variations. For example, we have a dataset of salaries, where each row is an employee and every employee record has 4 columns: age, number of years in a company, salary, bonus.

Sample — each row a dataset, in our case it’s employee. It can also be called as observation, example, instance, record.

Feature — each column of a dataset, in our example it’s age or number of years in a company or a salary. Also known as predictor, attribute, independent variable, input, regressor, covariate.

Label — is a value we’re predicting. In our example it can be a bonus, if we’re going to predict a bonus amount based on other features (age, years within a company, salary). Label also known as: target, outcome, response, dependent variable.