An early indicator to predict requirements volatility

Most development life cycle methodologies in government and industry follow some basic variation of the waterfall model. Volatility of requirements is a major concern with this life cycle model since the emphasis is on fully elaborated specifications as completion criteria for early requirements and design phases. This dissertation develops a methodology, termed VRAM or the Volatility of Requirements Assessment Method, to provide an early indicator of the volatility of requirements. The goal of this indicator was to enable the requirement engineer with a means to decide whether to move forward in the development life cycle or to engage in further requirements analysis based on the level of requirements volatility.; Based on a definition of requirements volatility and a literature review, 10 volatility factors were proposed. An experimental investigation determined a set of 7 empirically determined volatility factors that were employed by VRAM. The experimental investigation further compares an evaluation of requirements made by domain experts, using the proposed Volatility of Requirements Assessment Method, to assessments made by other subject matter experts, using their own domain expertise and judgment independent of VRAM. The assessments of both these groups of experts are evaluated against an actual database of requirement changes.; The results of the experimental investigation did not provide sufficient support to demonstrate that Subject Matter Experts using VRAM could effectively predict requirements volatility or its causes. Enhancements to the VRAM method were suggested and a recommendation made that future studies should be performed regarding the deployment of VRAM in the requirements engineering life cycle. The research concluded that not even experts were able to accurately predict volatility, emphasizing the necessary importance of future research in several areas. Among these are enhanced support aids to requirements engineers in assessing volatility and work involving development life cycles and combinations of life cycle models.