Equivalence Partitioning

Imagine you are writing a test for a simple age-based discount rule: under 5 is free, 5–60 is full price, over 60 is half-price. If you check ages 3 and 45 and both work, you might assume every age works—but the rule changes at 5 and 60. These change-points are called boundary points. Boundary Value Analysis is a testing method that checks the inputs exactly at these limits and just inside or outside them to catch errors where behavior shifts. Equivalence Partitioning complements this by letting you group ages into ranges where the behavior stays the same, so you test one sample per range plus those critical boundary points. This way you replace checking 120 possible ages with just a few smart choices.

Consider the same train ticket function in India. The input is age. The rule: children under 5 travel free; people aged 5 to 60 pay full fare; senior citizens over 60 get a 50% discount. Without a method you might pick ages 3, 27, 45, and 72. That misses what happens at the turning points: age 5 exactly, age 60 exactly, age 0, age 120. The function changes its logic at those edges.

Group the ages into classes where the outcome is identical:
- Under 5: free ticket. Sample: age 3. Edge tests: age 4 (last free), age 5 (first full).
- 5 to 60: full fare. Sample: age 30 or 45. Edge tests: age 5 (first full), age 60 (last full).
- Over 60: half-price. Sample: age 72. Edge tests: age 61 (first half), age 60 (last full).

If the code handles age 30 correctly, it will almost certainly handle age 45 correctly because both are in the same class with no logic change between them. Three or four well-chosen tests cover the behavior that would otherwise require checking 120 ages. The coverage is nearly the same as testing every value; only the boundary points carry the real risk, and Boundary Value Analysis focuses there.

A customs officer at an airport sorts passengers into groups like diplomats, crew, business-class travelers, and economy passengers with no checked bags, then inspects one representative per group. The group acts as a stand-in for everyone inside it because the processing rules are the same within each group; testing one reveals how all behave. A short caveat: this only works when the groups truly share the same handling logic, which in software means relying on the function's rules rather than visible surface traits.

Equivalence Partitioning is a test design technique that splits all possible inputs into ranges where the system behaves the same, then picks one sample from each range to represent it. This lets you replace many tests with a small set of representative values while still covering every distinct behavior.

It sits alongside Boundary Value Analysis (testing the exact limits and just-inside/outside points where rules change) and Pairwise Testing (combining choices from multiple input ranges instead of testing each alone).