Evaluation and Analysis of Applying Design Patterns

Abstract

Since applying design patterns provides a foresight of the exibility and future extensibility of software, their application is expected to constructing evolutionable software. However, there have been few research on quantitative evaluation of the e ectiveness of applying design patterns to software development. In this paper, we aim to evaluate the e ectiveness of applying design patterns. The C&K metrics suite is used for the evaluation. We collected C&K metrics values for two kinds of applications designed and implemented by our research group. Each application had two releases. One was a prototype release designed without using design patterns while the other was designed using design patterns. We conducted the Mann-Whitny U-test, one of the nonparametric statistics for testing hypotheses about whether two sample values di er. As a result, we found no signi cant di erences between the C&K metrics values of the two releases. Further examination of the relationship between the design patterns and the C&K metrics suite showed that speci c design patterns tend to make a particular metric value worse. The result suggests that new metrics should be devised for the evaluation of the e ectiveness of applying design patterns.