Software Delivery Risk Management: Application of Bayesian Networks in Agile Software Development

Abstract

-The information technology industry cannot be imagined without large-or small-scale projects. They are implemented to develop systems enabling key business processes and improving performance and enterprise resource management. However, projects often experience various difficulties during their execution. These problems are usually related to the three objectives of the project -costs, quality and deadline. A way these challenges can be solved is project risk management. However, not always the main problems and their influencing factors can be easily identified. Usually there is a need for a more profound analysis of the problem situation. In this paper, we propose the use of a Bayesian Network concept for quantitative risk management in agile projects. The Bayesian Network is explored using a case study focusing on a project that faces difficulties during the software delivery process. We explain why an agile risk analysis is needed and assess the potential risk factors, which may occur during the project. Thereafter, we design the Bayesian Network to capture the actual problem situation and make suggestions how to improve the delivery process based on the measures to be taken to reduce the occurrence of project risks.

Environmental software development in academic settings to support digital game-based learning: A case study on PEWI

“…The vast majority of these tasks were definable, however, with sufficient detail within the first three weeks of the summer. Fleshing out these earlier would have clarified the project's direction for all team members, mitigated schedule risk by adding increased clarity and details to schedule planning, and reduced stress on the project manager (Ancveire et al, 2015;Rose, 2013). We recommend "when in doubt, attempt to plan it out" as a good rule of thumb.…”

Using Bayesian Networks to Support Managing Technological Risk on Software Projects

“…Studies have applied Bayesian networks for supporting risk management for many specific purposes including setting a schedule [25]; budget definition [18]; quality assessment [1]; and defect prediction [20]. Fenton et al [9] created a Bayesian Network for risk management that supports decision making in various software design activities.…”

“…Khodakarami and Abdi [18] used a real example lived by one of the authors. Such validation strategy was also applied by Ancveire [1]. Fenton et al [9] made a simulation with two examples of projects to compare the view of experts with the results of the Bayesian network.…”

“…These studies have shown promising results, but they do not focus on risk factors related to the technological aspects of software projects. The risk factors used in their approach include information on requirements, human resources, management knowledge, quality process, and deliveries [1,9,18,25]. Some researchers investigate risk factors related to technological characteristics but simplify the analysis by defining only one risk factor to deal with all the project's complexity related to technologies [5,26].…”

“…Considering our proposed BN as a framework, the Core fragment would be the framework's cold spot. The Supplementary Material presents the definition for each Core node 1 .…”