Technical debt prioritization using predictive analytics

Recently, Technical Debt (TD) has gained popularity in the Software Engineering community to describe design decisions that allow software development teams to achieve short term benefits such as expedited release of code. Technical debt accrued should be managed to avoid the disastrous consequences of these temporary workarounds. Management of technical debt involve documenting the debt item in the backlog including some type of quantification in terms of person-hours or story points for example. Subsequently, the debt items are prioritized and addressed. Developers or project managers face problems to decide which debt is higher priority or more "problematic" and which one needs to be addressed first. This decision-making process is not standardized and is currently context dependent in most organizations. This paper bridge this gap by proposing a framework which makes use of a plethora of techniques ranging from data mining to prediction and decision models that project managers can use in their decision-making process to determine which technical debt is more critical and should be addressed first.