Towards a Bayesian Decision Model for Release Planning in Incremental Development
Olawole Oni · 2017
Incremental software development focuses on delivering working software in small increments so as to deliver early business value and minimize the risks of developing inadequate system requirements. Release planning is the activity that consist in planning what features and software qualities will be delivered in each release. Release planning decisions are complex due to conflicting stakeholders' objectives, uncertainty, and complex project constraints. Judgement-based approaches to release planning decisions are commonly adopted in practice but are subject to severe biases and lack of transparency. Model-based approaches have been proposed to address these problems but most methods use generic models that do not fit stakeholders' concerns and provide limited support for modelling and reasoning about uncertainty. The objective of this research is to propose a Bayesian approach to software release planning that uses economic and problem-specific, falsifiable models to support release planning decisions under uncertainty. Our approach uses Monte-Carlo simulation and multi-objective optimization to identify release plans that maximize net present value and minimize risks.