Mixed Integer Programming helping Requirements Allocation for the NRP in SCRUM Teams

Abstract

This paper proposes an approach based on mathematical modeling to provide decision making support for the Next Release Problem (NRP) arising in software development and maintenance companies that use SCRUM to manage teams and new releases of its software products. This approach aims to support the decisions from Product Owners and Teams about which requirements should be implemented for the next software release. Therefore it is proposed to use mixed integer programming in three different objective functions: (i) maximization of the customer satisfaction (considering their business value), (ii) minimization of software development time and (iii) minimization of the costs to produce the requirements. We consider constraints such as team availability, dependencies among requirements, the team's performance among others. The modeling was performed with the help of the AIMMS tool and some case studies were performed in three software development and maintenance companies using real data. The results show that it was possible to adapt the NRP for this kind of companies, achieving a better customer's satisfaction and business results using our approach.