A New Method for Software Size Estimation based on UML Metrics

In this paper we present a new method to estimate the size of a software project based on object-oriented paradigm. We want to calculate the complexity of a system from the most important features of the system that is described through UML diagrams. Our results show that the software size in the term of Source Lines of Code (SLOC) is well correlated with the complexity of classes which is computed from class diagram and sequence diagram. We also demonstrate that the Use Case diagram (UC) in the term of Unadjusted Use Case Point (UUCP) is a possible indicator for software size complexity. We use regression based model to build size estimation from complexity design measures and is concluded that some of metrics such as attribute, method, inheritance and communication between classes with regard to use case scenario of each class has fairly accurate prediction for size estimation (MMREs below 12%).

Keywords:

UML Metrics , Software Size Estimation , Unadjusted Use Case Point , Sequence , Class , Use Case Diagram.