Constructing C++ software size estimation model from class diagram

Software size is one of the most important internal attributes of software product. The information obtained from estimating software size will be useful for planning about effort, cost and activities schedule of later phases. This paper proposes an approach for constructing C++ software size estimation model using a statistical technique called Multiple linear regression analysis. The proposed model is constructed from structural complexity metrics that can be measured from class diagram. The paper also presents an automated tool for measuring these metrics, and in effect, estimating the C++ software size.