An evaluation of some design metrics

Estimating the quality of functional programs: an empirical investigation

“…As mentioned earlier, similar experiments to these have been performed for imperative languages within di erent application domains 3,6,7]. The programme of work described here has been in uenced by these experiments, particularly those partially funded by the Alvey Software Reliability M o d e l l i n g Project and the ESPRIT REQUEST project 7].…”

“…Metrics based on internal attributes are often employed in the belief that there is a strong correlation between internal attributes and the desired external attributes. Internal attributes such as length, modularity, reuse, coupling and cohesion are much easier to measure than external attributes, and some work has already been done on correlating internal and external attributes for programs written in imperative languages 3,6,7]. The experiment described here used established statistical techniques to ascertain whether or not there is any correlation between a selection of internal attributes (based on length and modularity) and certain characteristics of the development process, such a s t h e n umber of errors found during development, which are assumed to be indicative o f q u a l i t y.…”

“…The image analysis domain was chosen 3 because it o ers a particularly wide variety of distinct problems for software engineering and we already had expertise in the development of image processing algorithms in C 8]. The subset of algorithms examined so far has included low level algorithms which operate on an image array and produce image arrays as output (for example, convolution algorithms) and intermediate level algorithms which operate on image arrays and produce symbolic output (for example, curvilinear feature extraction algorithms).As mentioned earlier, similar experiments to these have been performed for imperative languages within di erent application domains 3,6,7]. The programme of work described here has been in uenced by these experiments, particularly those partially funded by the Alvey Software Reliability M o d e l l i n g Project and the ESPRIT REQUEST project 7].…”

D<inf>n</inf>-based architecture assessment of Java Open Source software systems

“…Assessing distribution by means of one of the statistical deviation values, e.g., the standard deviation, should be considered as one of the alternatives. However, the metrics are usually distributed in a highly asymmetric way [2,8] making standard deviation ill-suited for the distribution assessment. Therefore, one should estimate the probability density function for the distribution of the metrics being studied.…”

Predicting quality attributes of software product lines using software and network measures and sampling

“…Kitchenham et al showed that a useful prediction system can be created using statistical methods [7]. Following their example, we use a statistical method for discovering relations between internal and external attributes (Section 3.2).…”