It is more often used for descriptive or informative purposes. Its goal is to bring together observations according to pre-defined criteria by project teams and a predetermined number of groups. In this type of machine learning, the algorithm must group unclassed elements into groups according to their characteristics. Thanks to the increase in available data online, groups can be decided according to many more criteria.
For example, client bases can be segmented according to a larger number of criteria than before, and segmented more precisely – e.g., by industry, job type, seniority, hierarchy, location, etc. Once variables are selected and the number of groups is decided, the algorithm creates a series of rules to assign each individual to a group.
Learn more in our guide to machine learning.