Evaluating Design Metrics on Large-Scale Software

The purpose of the Design Metrics project is to develop a metrics approach for analyzing software designs which helps designers engineer quality into the design product. These metrics will gauge project quality as well as design complexity at all times during the design phase. Having quantifiable measurements could help managers and software developers determine the better design when alternative choices exist, as well as identify stress points which may lead to difficulty during coding and maintenance.

We have developed, for a structured design G, a design quality metric D(G) of the form D(G)=k₁D_e+k₂D_i. In this equation, k₁ and k₂ are constants and D_e and D_i are, respectively, an external and internal design quality component. In D_e we consider a module's external relationships to other modules in the software system, whereas in D_i we consider factors related to the internal structure. To form D_e and D_i, we searched for a combination of primitive design metrics which are useful, predictive, objective and automatable. This report will present our D(G), with its current D_e and D_i composite metrics, and empirical results as to how D(G) can identify stress points in a large-scale software design and how it is related to the quality of the resulting software.

    Purdue University
    1398 Computer Sciences Bldg
    West Lafayette, IN 47907-1398
    (765) 494-9329
    

Last modified 04/02/98 by Eric M. Bemus.

© Copyright, 1998, Design Metrics Research Team, Ball State University, All Rights Reserved.