Predicting change prone class in software is a difficult software engineering process. In order to keep up with the pace of modern day expansion, change in any software is inevitable software enhancement and defects are two main reasons for software change .In the earlier research, of predicting change prone classes the stress was given only on static characteristics rather than dynamic characteristics. In this paper, the authors aim to predict change prone classes using execution time of methods, trace events and dependencies between classes. Software prediction models based on these results can help them identify change prone classes of software which would lead to more rigorous testing and better results. They evaluate dependency between classes using UML2.0 sequence and class diagrams.