Optimization of Continuous Resource Allocation in Agile Scrum Software Development

Abstract

Scrum methodology is an agile software development process that breaks down complex problems and projects into feasible tasks that are methodically completed by a development team. It is led by a scrum master, who organizes the team and assigns these smaller tasks. Scrum simpli es software development into sprints of a set time frame. In the beginning, each team member is assigned tasks based on his skill set and how many story points he has available for that sprint. There are time inefficiencies in this process due to the need for manual sorting and assignment of the product backlog tasks to the team members during each sprint. In addition to initial assignments, issues arise when team members either over commit or under commit to a list of tasks at the beginning and need to adjust their task load before the sprint has ended. Un finishable tasks must be reassigned to team members who have additional available resources during the sprint as all tasks assigned at the beginning of the sprint should be finished by the end. In this thesis, the process by which these tasks are assigned to team members is analyzed to determine the factors to consider when allocating them. A variety of task assignment methods are tested using simulated story creation and completion to model the progress of a scrum development team. Speci fically, the business value-add and quantity of story points completed are analyzed and compared to determine the optimal allocation procedure. These methods are tested in systems with varying models of how business value is calculated. Simulations are also used to show the difference in the algorithm's effectiveness when the completion of tasks is guaranteed or variable. The ratio of business value-add to story point value is found to be the optimal method for task allocation with the method potentially changing slightly depending on how business value-add is calculated. This optimized assignment strategy has the possibility to drastically increase the sustained productivity of a development team by automating and maximizing the business value output of the scrum process.