Background. In the early phases of software development projects, thorough application of the COSMIC functional size measurement method may require more time and effort than available. Thus, early approximate methods have been proposed for estimating the COSMIC functional size of an application, instead of measuring it.Objective. The goal of this paper is to empirically evaluate the accuracy of the COSMIC early size estimation methods that are based on evaluations at the functional process level, for which historical data are available. The goal is to provide practitioners with empirical evidence on the accuracy of these methods.Method. We evaluated the Average Functional Process and the Equal Size Bands methods. We also proposed and evaluated two new approaches for defining bands in the Fixed Size Classification method. The estimation was performed by applying these methods to a set of software applications for which the data necessary to perform estimations were available, having been previously measured according to the standard COSMIC method.Results. Our analyses show that the Average Functional Process method generally provides estimates that are reasonable for early and quick sizing, but in some cases its estimation errors are too large to be acceptable. On the contrary, the methods using bands can provide quite accurate estimates. We determine the level of accuracy that can be obtained based on the type of method used, the number of bands used, and the quantitative characterization of the ability to classify each functional process in the correct band.Conclusions. The Average Functional Process method may be unreliable, as it occasionally yields quite large errors. Organizations using bands-based methods cannot just follow the prescribed estimation process: they need to properly train people in charge of classifying functional processes in the correct size band.

Empirical evaluation and proposals for bands-based COSMIC early estimation methods

Lavazza, Luigi;Morasca, Sandro
2019-01-01

Abstract

Background. In the early phases of software development projects, thorough application of the COSMIC functional size measurement method may require more time and effort than available. Thus, early approximate methods have been proposed for estimating the COSMIC functional size of an application, instead of measuring it.Objective. The goal of this paper is to empirically evaluate the accuracy of the COSMIC early size estimation methods that are based on evaluations at the functional process level, for which historical data are available. The goal is to provide practitioners with empirical evidence on the accuracy of these methods.Method. We evaluated the Average Functional Process and the Equal Size Bands methods. We also proposed and evaluated two new approaches for defining bands in the Fixed Size Classification method. The estimation was performed by applying these methods to a set of software applications for which the data necessary to perform estimations were available, having been previously measured according to the standard COSMIC method.Results. Our analyses show that the Average Functional Process method generally provides estimates that are reasonable for early and quick sizing, but in some cases its estimation errors are too large to be acceptable. On the contrary, the methods using bands can provide quite accurate estimates. We determine the level of accuracy that can be obtained based on the type of method used, the number of bands used, and the quantitative characterization of the ability to classify each functional process in the correct band.Conclusions. The Average Functional Process method may be unreliable, as it occasionally yields quite large errors. Organizations using bands-based methods cannot just follow the prescribed estimation process: they need to properly train people in charge of classifying functional processes in the correct size band.
2019
http://www.elsevier.com/wps/find/journaldescription.cws_home/525444/description#description
COSMIC measurement method; Early approximate methods; Equal size bands; Functional size measurement; Size estimation; Software; Information Systems; Computer Science Applications1707 Computer Vision and Pattern Recognition
Lavazza, Luigi; Morasca, Sandro
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11383/2077649
 Attenzione

L'Ateneo sottopone a validazione solo i file PDF allegati

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 6
  • ???jsp.display-item.citation.isi??? 4
social impact