AGILE SOFTWARE DEVELOPMENT RISK ASSESSMENT: A MONTE CARLO-BASED SIMULATION METHOD USING JIRA SCRUM AND KANBAN DATASETS

Abstract

Agile Software Development (ASD) has turned out to be the foundation of the present-day software development practices with the provision of iterative processes, such as Scrum or Kanban, that emulate focus on flexibility, cooperation, and continuous deployment. Nonetheless, it is always a cumulative issue to deliver on schedule when uncertainty strikes because the teams have changing loads, backlogs, and lack visibility on potential risks. Common planning methods, like average-based velocity or fixed sprint length, are unable to reflect real variation of the world and thus are subject to poor forecasts and lack of confidence by the stakeholders. To overcome this, the paper has suggested simulation-based risk analysis model which will rely on Monte Carlo Simulation (MCS) and will use the real empirical data of three practical software projects including GFG (Scrum), and User grid and Aurora (using Kanban). The time a particular historical issue was resolved and data on sprint performance were processed in order to calculate a velocity distribution which was in turn applied to predict the backlog completion timeline over 10,000 MCS iterations. Further improvement of simulations with SimPy further decreased the variability of the forecast and produced single estimate of project completion with a median (P50) of 3.45 sprints with a narrow 95% confidence interval of 3.45 to 4.60 sprints as opposed to the broader estimate of 16.67 sprints of the baseline MCS. Histograms, burn-up charts, and cumulative distribution graphs along with other visual solutions, ensured smooth intuitive performance of risk communication and comparisons between Scrum and Kanban. The framework also enables probabilistic planning (versus a static one) in as far as uncertainty is measured in the form of percentile-based estimates (as P10, P50, P90). The use of accessible technologies like Python and Jira REST APIs ensures the framework remains cost-effective and practical for Agile teams. This research advances Agile risk assessment by integrating statistical modeling with real project data, improving forecasting accuracy, stakeholder alignment, and delivery confidence—ultimately providing a scalable solution for managing delivery uncertainty in dynamic Agile environments.