Exploiting prior-phase effort data to estimate the effort for the subsequent phases
Filomena Ferrucci, Carmine Gravino, Federica Sarro · 2014
Context. Development effort estimation is a managerial activity that takes place throughout the life-cycle of the software so that it may benefit from information that becomes available as the project progresses. Researchers investigated the use of prior-phase effort data to estimate the effort in subsequent phases, as well as early phase effort data to estimate the total development effort. Objective. We assessed the usefulness of the effort spent for each phase in order to predict the effort required during the subsequent phase and until the end of the development process. We compared the use of effort data against the use of Function Points (i.e., a functional size measure widely used for effort estimation) and verified whether it is useful to combine them. Method. We performed an empirical study employing 25 applications from a single software company. The company collected effort from 3 different phases (i.e., specification and analysis, system and object design, and implementation and testing). Linear regression was used to build the estimation models. Results. Our analysis revealed that we obtained more accurate estimations by using prior-phase effort data to estimate the effort of subsequent phases. The combination of the prior-phase efforts and Function Points allowed us to improve the estimations in some cases. Conclusion. The effort spent in the prior-phase of a project is a good predictor for the effort that will be required later. In a continuous estimation process project managers can benefit from this effort data to obtain more accurate estimations for the subsequent phase(s).