Visualizes test ideas

The Classification Tree Method (CTM) visualizes test ideas. Because of the graphical description of the test case specifications, the intended tests can easily be reviewed, either by the developer of the test case specifications himself or by others (e.g. colleagues, certification institutes). Such a review builds up confidence in the tests and the appraisals of the user of the method and assures that no relevant test case has been overlooked.

Addresses complexity

If someone thinks about tests for a given problem, a lot of tests and testing ideas spring to one's mind. The challenge is more to structure the test ideas, to avoid redundancy (i.e. having several tests that test the same functionality) and to minimize the number of tests by avoid tests that test only one function instead of several functionalities, if this is possible.

In that, the systematic approach of the CTM is very useful. By dividing the value ranges in classes, it becomes clear, how many values of a test-relevant aspect have to be used. (E.g. it is not useful to introduce additional tests only to test different values from one class. If this is necessary, this class should have been divided into subclasses in the tree.)

Checks Specification

A major task when applying the Classification Tree Method is to analyze the specification of the functional problem, e.g. when determining the test-relevant aspects of the problem. In doing so, the specification is checked exhaustive for consistency, omissions, etc. E.g. if a expected behavior for a given test case cannot be determined from the specification, there is obviously something missing in the specification.

Measures Test Expenditure

The Classification tree allows to calculate some metrics, e.g. the minimal and maximal number of test cases needed for a given tree. This allows to estimate the required test effort at an early state in the development process.

Builds up confidence

Having a well-designed tree and well-thought test case specifications combined from its classes, gives high probability that this set of test cases is sufficient for the problem in question, i.e. these tests will detect practically "all" errors. This holds even more true if the test case specifications have gone through one or more review processes.

Test End Criterium

With such a confidence, it is easy to determine when it is safe to end the tests, because it is simply the point in time, when all tests pass. If no relayed-upon set of test cases exists, there is always uncertainty if one should not do another additional test.

A relayed-upon set of test cases allows for refracturing of the test object, i.e. changing of the internals of the test object for reasons of improvement without having to worry about the required tests to re-check the (external) functionality.