Evolving estimators for software project development

Abstract

In this research, an application of a computational intelligence approach for effort estimation in software projects is presented. More specifically, the authors examine a genetic programming system for symbolic regression; the main goal is to derive equations for estimating the development effort that are highly accurate. These mathematical formulas are expected to reveal relationships between the available input features and the estimated project work. The application of the proposed methodology is performed in two software engineering domains. The proposed model is shown capable to produce short and handy formulas that are more precise than the existent in literature. © 2011 Springer-Verlag.