Machine learning (ML) is a collection of data-driven algorithms that work without being explicitly programmed for a specific task. Unlike non-ML algorithms, they require (and learn by) the training data. ML algorithms are classified into supervised and unsupervised types.

Supervised learning means that the training data consists of input vectors and their corresponding output value as well. This means that the task is to establish relationships between inputs and outputs in a historical database, called the training set, and thus make it possible to predict outputs for future input values.

For example, banks have vast databases on previous loan transaction details. The input vector is comprised of personal information—such as age, salary, marital status and so on—while the output (target) variable shows whether the payment deadlines were kept or not. In this case, a supervised algorithm may detect different groups of people who may be prone to not being able...