Software projects have a high risk of cost and schedule overruns. The ability to make reliable predictions of development effort and delay risks would help software companies to improve customer satisfaction, while reducing costs and optimizing delivery speed. Nowadays most software companies operate in highly dynamic environments that require software development teams to respond fast to change and continuously deliver business value. These circumstances call for novel estimation methods that are able to continuously learn and adapt to changes in the environment of modern software development teams.
Machine learning models have recently gained popularity in the areas of effort estimation and risk prediction. Recent work used deep learning to estimate the size of user stories through learning a team’s previous estimates. Other work focused on predicting delay risks and sprint delivery risks. Predicting risks requires the capability of processing large amounts of historical project data, using dynamic and contextual features, and tracking dependencies between (subsets of) development tasks.
The paper to be discussed in this session is: [1].
The course contents are copyrighted (c) 2018,2019,2020 - onwards by TU Delft and their respective authors and licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International license.