What Is Usage-Based Statistical Testing?

by Nataliia Vasylyna | January 20, 2012 1:30 pm

Actual client usage of software may be considered as a form of usage-based testing. In the case if defects are detected by clients, some information about them can be reported to software vendors, and integrated fixes may be created and delivered to all the clients to avert such defects.

Nevertheless, correcting software bugs after release could be very costly because of the massive quantities of software installations. Frequent fixes could also harm the software vendor’s reputation and long-term business vitality.

If the actual usage or expected usage for a new product can be captured and used in software testing, product reliability could be most straightly guaranteed.

In usage-based statistical testing the testing environment resembles the actual operational environment for the software in the field, and the overall testing sequence, as represented by the orderly implementation of specific test cases in a test suite, resembles the usage scenarios, sequences, and templates of actual software usage by the target clients.

Because the huge quantity of clients and diverse usage templates cannot be captured in an implementation set of test cases, statistical sampling is required, that is why we use the term “statistical” in this strategy. We can also use the term “random testing”.

Usage-based statistical testing is commonly appropriate to the final phase of software testing, normally referred to as acceptance testing[1] right before product release, so that stopping testing is of equal worth to the product release.

More lately sub-phases of testing, such as integration and system testing, could also favor the knowledge of actual client usage situations to drive effectual reliability betterment before product release. Certainly, the termination criterion used to stop such testing is fulfillment of reliability purposes.