Towards making functional size measurement easily usable in practice

Functional Size Measurement methods –like the IFPUG Function Point Analysis and COSMIC methods– are widely used to quantify the size of applications. However, the measurement process is often too long or too expensive, or it requires more knowledge than available when development effort estimates are due. To overcome these problems, simplified measurement methods have been proposed. This research explores easily usable functional size measurement method, aiming to improve efficiency, reduce difficulty and cost, and make functional size measurement widely adopted in practice. The first stage of the research involved the study of functional size measurement methods (in particular Function Point Analysis and COSMIC), simplified methods, and measurement based on measurement-oriented models. Then, we modeled a set of applications in a measurement-oriented way, and obtained UML models suitable for functional size measurement. From these UML models we derived both functional size measures and object-oriented measures. Using these measures it was possible to: 1) Evaluate existing simplified functional size measurement methods and derive our own simplified model. 2) Explore whether simplified method can be used in various stages of modeling and evaluate their accuracy. 3) Analyze the relationship between functional size measures and object oriented measures. In addition, the conversion between FPA and COSMIC was studied as an alternative simplified functional size measurement process. Our research revealed that: 1) In general it is possible to size software via simplified measurement processes with acceptable accuracy. In particular, the simplification of the measurement process allows the measurer to skip the function weighting phases, which are usually expensive, since they require a thorough analysis of the details of both data and operations. The models obtained from out dataset yielded results that are similar to those reported in the literature. All simplified measurement methods that use predefined weights for all the transaction and data types identified in Function Point Analysis provided similar results, characterized by acceptable accuracy. On the contrary, methods that rely on just one of the elements that contribute to functional size tend to be quite inaccurate. In general, different methods showed different accuracy for Real-Time and non Real-Time applications. 2) It is possible to write progressively more detailed and complete UML models of user requirements that provide the data required by the simplified COSMIC methods. These models yield progressively more accurate measures of the modeled software. Initial measures are based on simple models and are obtained quickly and with little effort. As V models grow in completeness and detail, the measures increase their accuracy. Developers that use UML for requirements modeling can obtain early estimates of the applications‘ sizes at the beginning of the development process, when only very simple UML models have been built for the applications, and can obtain increasingly more accurate size estimates while the knowledge of the products increases and UML models are refined accordingly. 3) Both Function Point Analysis and COSMIC functional size measures appear correlated to object-oriented measures. In particular, associations with basic object- oriented measures were found: Function Points appear associated with the number of classes, the number of attributes and the number of methods; CFP appear associated with the number of attributes. This result suggests that even a very basic UML model, like a class diagram, can support size measures that appear equivalent to functional size measures (which are much harder to obtain). Actually, object-oriented measures can be obtained automatically from models, thus dramatically decreasing the measurement effort, in comparison with functional size measurement. In addition, we proposed conversion method between Function Points and COSMIC based on analytical criteria. Our research has expanded the knowledge on how to simplify the methods for measuring the functional size of the software, i.e., the measure of functional user requirements. Basides providing information immediately usable by developers, the researchalso presents examples of analysis that can be replicated by other researchers, to increase the reliability and generality of the results.