To reduce the development cost and time that are required to perform a software project, software organizations make an effort to predict exactly and to improve steadily the quality of the software produced and the productivity achieved in the project. These quality and productivity are influenced by the processes used for executing a software project. Therefore, software organizations have to improve software processes by making the knowledge gained from past projects persistent and thus reusable for future projects. In order to manipulate the knowledge obtained from projects, we use analysis methods of project processes. Since most of existing analysis methods use the characteristics of software products as the analyzing factors, they do not provide the analysis method with respect to process components such as activity, artifact, and agent. In order to provide the analysis information on process component-level such as the dependence relations and precedence relations among activities and among artifacts, we propose analysis methods from the viewpoint of processes and a process analysis framework.
In this thesis, we develop a Model-based Process Analysis Framework(MPAF) in order to improve the software processes and conduct an empirical analysis with 10 projects. Since it is not easy to analyze raw project execution data in artifacts from the point of processes, a process model is used in MPAF for abstracting and understanding of the process and for supporting of these analysis. MPAF provides the analysis components and the procedure that constructs the process model instance using information extracted from executed projects and analyzes it using the defined analysis factors with the viewpoint of processes. We can analyze the complex processes used for executing a software project with high understanding degree, support to find easily the point of improvement at process component-level, and automate these analysis procedure in PSEE(Process-centered Software Engineering Environment) using MPAF.