Accurate effort estimation enables effective managerial decisions to be made by project managers when embarking on a project, and this trend also applies when managing Web projects. Web development has steadily increased over the years; research into identifying sound ways to improve the effort estimates being made by thousands of Web companies worldwide would be valuable. Looking at the state of the art in the domain of Web resource estimation, we show that numerous effort estimation techniques have been investigated; none of which have proven conclusively to be the best.I would like to thank my supervisors, Pat Riddle and Emilia Mendes, for their patience, guidance, and encouragement over the course of this PhD. It has been a long journey, and without their invaluable input and understanding, this thesis would not have been possible.I would like to thank the Department of Computer Science at the University of Auckland for giving me the opportunity to do a PhD here, for providing financial support in terms of my PhD scholarship and employment opportunities, as well as for all their technical and administrative help.I would like to thank my family for their patience, love, and support, through the various ups and downs natural to an endeavor as complicated and lengthy as this.Lastly, to my friends and colleagues, thank you for your support, academic and otherwise, that have made my PhD experience both interesting and enjoyable.
Damir Azhar31/08/2015iii iv
Contents
List of Figures ix
List of Tables xiiiClick hereClick here using Web project data from the Tukutuku dataset [4].3. We expanded on the replicated methodology, introducing the use of bootstrap aggregation, commonly referred to as bagging, to the process of ensemble creation. This too is novel in the area of Web effort estimation. Bagging enables multiple experimental runs to be performed using a single dataset. We investigated three variants of bagging using 10 experimental runs for each variant, to provide a clearer picture of ensemble Web effort estimation performance.4. Using ensemble results obtained from the bagging experiments, we broke down ensemble performance via a mathematical formalization of the accuracy-diversity trade-off. This allowed us to quantify the relationship between ensemble error, ensemble diversity, and the accuracy of the component classifiers. The insight obtained from the analysis of this relationship is valuable for creating effective ensembles.All of the above contributions collectively further the general understanding of ensemble behaviour and performance, when used for Web effort estimation. Practitioners will be able to use these contributions to implement effective effort estimation ensembles, while researchers can build on them to further research into this domain.
OrganizationThe remainder of the thesis is organized as follows:Chapter 2 presents a systematic literature review of Web resource estimation, detailing both the review process and its findings. We conclude this chapter with a discussion of what can be done to...
