Abstract:
Model-Driven Engineering (MDE) is used in the Software Industry which enables level to level transformation until the final system is created. This concept helps to ensure the bridging of gaps between the problem domain and the solution scope of a software system. A software system with a lesser number of software defects or zero defects will be successful software. Earlier the defects are identified it reduces the cost in terms of effort, time, and human resources, rather fixes that defect in a later stage of the software development life cycle. The development of defect prediction models and the efficient usage will prevent unnecessary defect fixing efforts later.
Unified Modelling Language (UML) provides certain notations to create models in different aspects. UML Class diagram is very widely used in identifying and evolving business entities. UML Class diagrams as entities can be mapped with Database Management Systems and propagate the business entities.
Every business must deal with the inevitable truth of change. To survive in a competitive market, business functions, and business directions are under the freedom of change at any moment. Stable business solutions are the compulsory components of successful businesses.
Applying changes to Software makes them fragile when they are not done properly. The phase where adding changes or in the maintenance mode, and the most important stage of the business, must be well away from defects.
This thesis covers a defect predictive approach that can be applied at the UML class diagram models that are created at the beginning of the Software solution, however, the defect prevention applies to the maintenance or in the most vital stage of the business.
This thesis discusses the possible defect prediction models that can be used in MDE to facilitate fast and efficient software development. At the end of the thesis, it will discuss the approach that has taken to introduce a defect prediction strategy to the Model-Driven Engineering its evaluation and the contributions to the research community
