Self-Organizing Maps for Agile Requirements Prioritization

In building software systems, decisions at the specification phase will extremely affect the rest of the system life cycle. Well-defined requirements at this phase will increase the chance of achieving the ultimate goal of delivering a software that meets stakeholders needs. Given a limited sources of time and predefined budget, not all the requirements should be fulfilled with the same priority. Here comes the need for requirement prioritization RP techniques. This paper presents a new approach to deal with the dynamic nature of requirements prioritization process in agile development. Training a self-organizing map according to requirement's predefined features is the main process in the proposed approach. The trained map can produce a set of clusters. A farther rank is given to each requirement according to map resulting weights. The proposed approach was implemented using different variables related to requirements themselves and related to the self-organizing map to show its ability to prioritize requirements in agile development model.