Optimal Sprint Length Determination for Agile-Based Software Development

Abstract

A carefully planned software development process helps in maintaining the quality of the software. In today's scenario the primitive software development models have been replaced by the Agile based models like SCRUM, KANBAN, LEAN, etc. Although, every framework has its own boon, the reason for widespread acceptance of the agile-based approach is its evolutionary nature that permits change in the path of software development. The development process occurs in iterative and incremental cycles called sprints. In SCRUM, which is one of the most widely used agile-based software development modeling framework; the sprint length is fixed throughout the process wherein; it is usually taken to be 1-4 weeks. But in practical application, the sprint length should be altered intuitively as per the requirement. To overcome this limitation, in this paper, a methodical work has been presented that determines the optimal sprint length based on two varied and yet connected attributes; the cost incurred and the work intensity required. The approach defines the number of tasks performed in each sprint along with the corresponding cost incurred in performing those tasks. Multi-attribute utility theory (MAUT), a multi-criterion decision making approach, has been utilized to find the required trade-off between two attributes under consideration. The proposed modeling framework has been validated using real life data set. With the use of the model, the optimal sprint for each sprint could be evaluated which was much shorter than the original length. Thus, the results obtained validate the proposal of a dynamic sprint length that can be determined before the start of each sprint. The structure would help in cost as well as time savings for a firm.

Maturing the Scrum Framework for Software Projects Portfolio Management: A Case Study-Oriented Methodology

“…This is due to the fact that there is no previous history with rich data associated with the upcoming subprojects of SCMP. Thus, the proposed heuristic score-based method outperforms [33], since it does not depend on the presence of a large amount of data.…”

“…An alternative maturing practice consists in tuning the settings of the Scrum framework. In this regard, [33] used a multi criteria decision making-based approach to optimize the sprint length of agile software development in terms of the underlying cost and work intensity of the planned work. However, their approach requires a cumulative distribution function related to the pattern of performing user stories, which makes it intractable for projects with lacking data.…”

“…Regarding, in particular, the proposed heuristic scorebased method for computing the sprint length, its superiority in comparison with the previous studies is in the availability of data. In SCMP, where new subprojects are born with an increasing trend, the sprint length calculation methods based on the statistical inferences ( [33]) are not workable. This is due to the fact that there is no previous history with rich data associated with the upcoming subprojects of SCMP.…”