Modelos segmentados de estimación del esfuerzo de desarrollo del SoftwareUn caso de estudio con la base de datos ISBSG
-
1
Universidad de Valladolid
info
-
2
University of Reading
info
-
3
Universidad de Alcalá
info
ISSN: 1886-4554
Any de publicació: 2004
Volum: 1
Número: 2
Pàgines: 25-30
Tipus: Article
Altres publicacions en: Revista de procesos y métricas de las tecnologías de la información
Resum
Parametric sofware effort estimation models use historical project databases to adjust the parameters of the required effort function. The use of databases with data coming from heterogeneous sources often entail that the resulting models are subject to excessively high mean errors, due to the fact that data are widely diverging in magnitude. In order to improve this situation, the segmentation of the input domain has been proposed, so that a regression model with different parameters is obtained for each segment. In this paper, the use of well-known clustering algorithms in a recursive way is described as a technique to obtain segmented models of the kind mentioned. Concretely, the ISBSG database and the EM algorithm are use as a demonstration of the results that can be obtained through that technique.