Using Web objects for estimating software development effort for Web applications

Abstract

Web development projects are certainly different from traditional software development projects and hence require differently tailored measures for accurate effort estimation. We investigate the suitability of a newly proposed size measure for Web development projects: Web objects. Web objects have been specifically developed for sizing Web applications and used for estimating effort in a COCOMO Il-like estimation model called WEBMO. However, no empirical validation has yet been published. We apply and validate the proposed Web object approach in the context of a small Australian Web development company, for the first time. Besides Web objects, we apply traditional function points as an effort predictor for Web applications. Effort estimation models based on Web objects are compared with models based on traditional function points using ordinary least squares regression (OLS). Tested on data from twelve Web applications, the estimates derived from estimation models using Web objects significantly outperformed models using function points, with a mean magnitude of relative error of 0.24 versus 0.33, respectively. Based on the results, it seems that Web objects are more suitable for effort estimation purposes of Web applications than traditional function points.