Evaluation of optimized staffing for feature development and bug fixing

Abstract

Skill level and productivity varies substantially between developers. In current staffing practices, however, developers are largely treated as the same. In this paper, an empirical analysis of the tow formulations of assignment of developers to tasks and bug fixing activities is studied. Two related problems are considered:(i) Assignment of developers to bug fixing with the objective to achieve best match between requested skill profile and assigned developer's skill profile.(ii) Assignment of developers to feature-related tasks in iterative development process.Two optimization approaches have been customized to determine qualified staffing plans. They are based on greedy optimization respectively genetic algorithm (GA). Empirical analysis is done for nine milestones of the open source Eclipse JDT project and two industrial case study projects. The main conclusion drawn from the analysis is that substantial savings can be achieved from optimized staffing policies when compared to the manual plans formerly applied. More specifically, the GA results are mostly the best, and the (lightweight) Greedy search becomes the better the bigger the look-ahead time L. Overall, the results are considered as decision support in finding better staffing policies in shorter time.

An empirical investigation of single‐objective and multiobjective evolutionary algorithms for developer's assignment to bugs

“…Before we proposed a GA‐based approach for assigning developers to bugs , the K‐greedy approach , a greedy‐based technique, was one of the three automatic approaches for assigning developers to bugs. Given a bug's arrival (filing) time, this algorithm considers all the developers who are available at and after that time (until a time window called the look‐ahead time (LAT)).…”

“…Rahman et al proposed a GA‐based approach for developer's assignment to bugs. Our bug assignment approach has some differences compared with their approach.…”

“…In our case, we order the bugs by using a topological sorting algorithm exploiting dependence among bugs. Rahman et al used different types of developer assignment techniques for different types of issues (e.g., bugs, features). Even though in this paper we deal only with bugs, the proposed individual competency model and the solution approach can be easily extended to other types of issues (e.g., new features, enhancements to existing features).…”

“…To reduce developers’ time and effort spent on bug triaging, over the years, researchers have developed various techniques based on machine learning (ML), information retrieval (IR), graph algorithms, and greedy algorithms for the bug recommendation purpose. In making recommendations, most of the bug triaging approaches (except ) do not take developer workload into account. A developer might be the best match for fixing a bug, but the developer might be overloaded with bugs.…”

Search-Based Software Project Management

“…Kang et al (2011) optimized the scheduling of human resource allocations by using a variant of SA, taking into account individual and team constraints based on the literature and interviews with experts in the industry, and by employing real data to validate their proposal. Rahman et al (2010) also reported on an empirical analysis carried out exploiting real data on the use of both GA and a greedy approach that makes the locally optimal choice at each stage to assign developer to tasks and bug fixings activities. Different aspects of the management also focused on the allocation of staff (Barreto et al 2008) (Kapur et al 2008) and the provision of decision support (Cortellesa et al 2008).…”