Study of metrics and practices for improving object oriented software quality

Modern software systems are large and complex products, consisting in thousands lines of code, developed, often in a distributed environment, by dozens of developers and produced through an industrial process, usu- ally with short time to market. To manage such kind of complexity and to keep the development process under control measurements and metrics are required. The present thesis collects the outcomes of the research the author carried on in the field of software metrics during the three years of the Ph.D. studies. Software metrics are used to measure various aspects of software development, including software features, processes execution, developers' efforts, software quality, just to name a few. The first part of the present thesis reports the results of the studies performed on product metrics, with the final goal of helping software engineers better manage the programmers efforts and particularly to assess software quality dur- ing software development. The second part of this dissertation presents the outcomes of the research aimed at shedding some light on the effec- tiveness and impact of some development practices on software systems. To perform these studies I used a novel approach, based on the concept of complex network. Complex networks are in fact one of the best can- didates to represent software systems, enabling researchers to obtain a deeper knowledge of the structure and evolution of a software system. We found some meaningful statistical correlations between network metrics and software properties. Both the theoretical framework and the reported findings might, in principle, have a practical application to assist software engineers dealing with specific development tasks, like bug discovery or refactoring.