UML Based Effort Estimation in Component Based Systems

Abstract

Software development has come a long way from traditional software development, which is characterized by the structured programming paradigm introduced in the late 60’s and early 70’s to contemporary development practices, which characterize a software application as interacting, independent components. Traditional software effort estimation models capture this monolithic view of software development. In these models, software effort is projected at the large-grained system level whereas to accurately predict effort in Component Based Software Development (CBSD), a fine-grained approach is needed to identify and classify the relevant cost factors. Effort estimation in CBSD is concerned with deriving estimations for small, concurrent development projects. The small projects allow for quantitative and direct measurements of the factors influencing cost. CBSD is principally based on the core concepts of Object Orientated Techniques where the role of UML is unbeatable. Different UML constructs can be used at different stages of software development for estimation of resources like efforts & cost etc. In this study, the literature was reviewed for various estimation techniques based on UML diagrams and their pros and cons in context of CBSD were underlined. The important aspect of CBSD is its functional and non-functional requirements from the integrated set of components. These were incorporated into newly defined set of technical and environmental factors and other major issues like use case granularity etc. were collected during the computations. Karner’s use case point method was adapted and the revised framework was applied on two case studies. Validation was done on another similar work done by Bente Anda et al. and the results varied just about 6.15% with respect to actual efforts made. This study thus, shows how complexity factors can be calibrated for a specific context and produce relatively accurate estimates.