Department of Computer Engineering, National Institute of Technology, India
Abstract: Data mining plays vital role in data analysis and also encompasses immense
potential of mining software engineering data to manage design and maintenance
issues. Change impact assessment is one of the crucial issues in software
maintenance. In Object Oriented (OO) software system, classes are the core
components and changes to the classes are always inevitable. So, OO software
system must support the expected changes. In this paper, to assess impact of change in the class, we have
proposed changeability measures by mining associations among the classes. These measures estimate a) change
propagation by identifying its ripple effect; b) change impact set of the
classes; c) changeability rank of the classes and d) class change cost.
Further, we have performed the empirically study and evaluation to analysis our
results. Our results indicate that by
mining associations among the classes, the development team can
effectively estimate the probable impact of the class change. These measures
can be very helpful to perform changes to the classes while maintaining the
software system.
Keywords: Mining software engineering data, object oriented system development,
change propagation, change impact.