This technique divides input data into data classes to reduce test cases amount to validate a rule. The idea is that given a set of possible equivalent values, using just one of those values will be enough to design a test case.
An advantage of this approach is reduction in the time required for testing a software due to lesser number of test cases.
It strives to find errors that may arise based on information classes and reduce to the minimum the effort needed in terms of test case design as well as test data mass.