Framework para estimar esforço de manutenção em um ambiente multi-equipe

Abstract

During the software lifecycle, good effort estimation allows teams to make decisions about how activities will progress. Inaccurate or incorrect estimates directly impact software delivery, customer dissatisfaction, and ultimately, project failure. Several approaches proposed methods to calculate effort estimation that is approximate to the real effort. However, there is a gap in the state of practice concerning effort estimation in software maintenance activities. This work proposes a framework to calculate effort estimates during software maintenance, in a company environment composed of multiple teams working on the same project. This framework uses the real effort from previous activities to estimate the effort for new ones. To validate the EstimAi framework, an experimental evaluation was conducted in two maintenance teams during two Sprint iterations. As a result, the approach is not accurate; however, according to developers’ feedback, the estimate was acceptable, which leads us to conclude that the framework needs adjustments to further improve usability and the generated estimate.