Which pertinent metrics used to evaluate the software maintainability ?

Meskens, Nadine [UCL] Philippard, David Levecq, Philippe [FUCAM]

(eng) The reduction of computer hardware costs has led to an increasing demand and multiplication of software programs. Unfortunately, software costs are not decreasing. Over 70 % of the software development effort is spent today in testing and maintaining existing software. The increase of the maintenance burden is due for the most part to the progressive degradation of the program's intrinsic quality. A first objective of our study is to select suitable metrics for analysing the maintainability of a program and to provide maintainers with a set of pertinent measures on the intrinsic quality of a program. The second objective is to detect the existence of a link between the classification of programs and their belonging to a group of renovation actions to be undertaken on them (the program has to be formatted, restructured, rewritten, divided up or documented). To define a set of pertinent metrics several methods can be used such as the discriminant analysis and the rough set theory.