A complexity measure during software design
J L Munoz · 1986
A multi-variate complexity metric was defined and applied to a collection of modules represented by both a design and an implementation of that design. The application of complexity to software designs provided results superior to their application to the code implemented from that design. It is suggested that this is a result of noise in the implemented code. In addition, a linear relationship was found to exist between the number of statements in a PDL representation of a design and its implemented counterpart. Finally, it was found that the individuals involved in the design and implementation phases must be factored into any complexity studies. A multi-variate approach addresses concerns related to situations where conflicts arise as a result of using more than one complexity measure. The multi-variate complexity metric defined contained both intra- and inter-module features. A statistical approach was applied to identify the significant features of the multi-variate.