Grip on software: understanding development progress of SCRUM sprints and backlogs

Helwerda, L.S. (2024)

Doctoral Thesis

Software development is a complex process. It is important that software products become stable and maintainable assets. We examine and extend a software development framework that has been designed to support development in a flexible manner, namely Scrum. We apply machine learning methods in the hope to better understand what happens during the development process and provide predictions of Scrum sprint progress and product backlog sizes, among other indications of success. To obtain our data set, we design and construct a data acquisition pipeline using agent-based networking and query templates for generalized feature extraction. We provide overviews and in-depth reports for decision support and highlights of other aspects of the development ecosystem through information visualizations.

• Software development process
• Scrum
• Data acquisition pipeline
• Pattern recognition
• Machine learning
• Feature extraction
• Product backlog estimation
• Story point estimation

• All authors: Helwerda, L.S.
• Supervisor: Verbeek, F.J.; Kosters, W.A.
• Committee: Bonsangue, M.M.; Kleijn, H.C.M.; Manegold, S.; Chaudron, M.R.V.; Soomlek, C.; Niessink, F.
• Qualification: Doctor (dr.)
• Awarding Institution: Leiden Institute of Advanced Computer Science (LIACS), Faculty of Science, Leiden University
• Date: 2024-09-13
• Title of host publication: SIKS Dissertation Series
• Name: 2024-28
• Sponsorship: Funded by Stichting ICTU as part of a collaboration between Leiden University and ICTU