A novel change impact model for enhancing project management

Abstract

Software systems changes constantly with time. Changing the software affects all the classes associated with it. For effective project management it becomes important to predict change impact classes in earlier phases of software life cycle. This paper aims to develop a novel model using dynamic metrics and several behavioural dependencies. Using code analyser trace events 30 different metrics are analysed which are further used for refining the degree of change impact feature of a class. Further the model is validated using K-means clustering technique, naive Bayes classification and logistic regression in WEKA tool. Validation of the model is done using open source software Art of Illusion (AoI).