Date of Award

2003

Document Type

Master Thesis

Degree Name

Masters of Science (Research)

Department

Department of Computing

First Advisor

Dr. Pat Doody

Abstract

This thesis looks at how the size of a software product can be measured. It looks at current software sizing methods (LOC and FPA) and shows that they are not sufficient in today’s rapidly changing environment. The COSMIC-FFP method is introduced and an overview of how it works is given. The hypothesis to be tested in this thesis is whether COSMIC-FFP is an accurate, repeatable sizing method that when applied to any type of software project produces a size measure that can be used to predict the effort required to complete the project.

This thesis investigates the use of the COSMIC-FFP method with past software design methods (Data Flow Diagrams), current design methods (Unified modelling Language) and future design methods (Agile Methodologies). It shows that there is a clear logical mapping between these methods and the COSMIC-FFP model. Mappings from these design methods to the COSMIC-FFP model have been developed and tested. Using these mappings to execute the size count allows for repeatability and accuracy. They also allow the sizing exercise of software projects to be easily automated by incorporating the count into design tools.

The size of a project is used to predict the effort required to complete a project. To test a sizing method one needs to check the relationship between size and effort. If a statistically significant relationship exists between these two variables then the sizing method can be used to predict effort. To prove the second part to the stated hypothesis an empirical evaluation of the size/effort relationship for projects that have been sized using the COSMIC-FFP method is carried out. The results obtained show that COSMIC-FFP can be applied to any type of software project and the size count can be reliably used to predict the effort required for the project.

This may be of importance to the software community. Having a reliable, accurate size measure for past, current and future projects of any type will allow organisations to compare projects and so enhance software process improvement.

Access Level

info:eu-repo/semantics/openAccess

Share

COinS