Application of a genetic algorithm to optimize staffing levels in software development

Abstract

Human resource management is gaining recognition as being one of the core attributes of effective software project management. With the great dependency the Department of Defense (DoD) has on systems software, the development of an optimal staffing policy that minimizes overall software project costs will be very valuable. This research aims at developing such a system using a dynamic simulation model that incorporates data collected during an actual software development project. A genetic algorithm is used to arrive at a solution to this nonlinear optimization problem. The simulation model accepts varying staffing schemes supplied from the genetic algorithm allowing for the examination of the effects of the staffing schemes on total project cost. The results indicate that the system was able to obtain a solution reducing overall project costs, but at the cost of additional workdays.