An analytical model for requirements selection quality evaluation in product software development

Abstract

In market-driven development of product software, a crucial decision for each candidate requirement is whether or not to select it for implementation in the next release. We present an analytical model of the requirements selection process, which is used for reasoning about decision quality. A network of queues, with two classes of jobs, models the selection of requirements of different quality. The feasibility of model parameter estimation is validated in a survey involving product managers and system engineers. The results show that some of the respondents have made internally consistent parameter estimations, indicating that the model is relevant and its parameters understandable. It is also shown that a majority of the consistent respondents estimate that most of their implemented product requirements are incorrectly selected. The main objective of the model is to provide tools for evaluation of improvement proposals by estimating the impact of process change.