A complexity metric for object-oriented software

With the continuous expansion of software scale, the complexity of software is getting increasing. The relationship between the models of large software projects is very complicated. In the development of software, how to make a reasonable evaluation of the quality of software projects has become a very important and urgent problem. Class diagram as a very important diagram in the process of software design, its reasonable design has a remarkable effect on the software complexity. To study the complexity of the class diagrams, this paper proposes maximum deviation-based multi-attribute decision method to measure class diagrams. By using the method of maximizing deviations, we can measure the complexity of the object-oriented software scientifically and reasonably. And then through case analysis, Pearson simple correlation coefficient was used to analyze the correlation between the existing three measurement methods and the metric of practical experience, whose results show that while measuring the object-oriented software, maximizing deviation method is more in accordance with practical experience. © 2018, © 2018 Informa UK Limited, trading as Taylor & Francis Group.