A generalized duration forecasting model of test‐and‐fix cycles

Rework is inherent in product development, but estimation of the duration and required resources for rework cycles is inhibited by task size variation, process capacity variation, and the process characteristics of dynamism and concurrency. This difficulty is especially evident in testing of software-intensive systems. Dynamic modeling of test-and-fix cycles has addressed this problem by providing good forecasting of test phase durations and by providing good results in supporting resource allocation and process improvement decisions. A generalized model of test-and-fix phases has been produced from an analysis of commonality and variation in six project models. The generalized model can be configured to support test phase planning of duration and resources. This paper extends the work reported in ICSSP 2012 by reporting results of experiments on managing test-and-fix projects through work-in-process limits and through test case prioritization. Copyright © 2014 John Wiley & Sons, Ltd.