Abstract The paper describes a method of software quality control based on the use of software metrics. The method is applied to software design metrics to illustrate how design metrics can be used constructively during the software production process. The various types of design metrics and how they can be used to support module (procedure) quality-control are discussed. This involves adapting conventional quality control methods such as control charts to the realities of software, by using: ‘robust’ summary statistics to construct ranges of acceptable metric values; scatterplots to detect modules with unusual combinations of metric values; and different types of metrics to help identify the underlying reasons for a module having unacceptable (anomalous) metric values. The approach is illustrated with examples of metrics from a number of existing software products.
[1]
Anas N. Al-Rabadi,et al.
A comparison of modified reconstructability analysis and Ashenhurst‐Curtis decomposition of Boolean functions
,
2004
.
[2]
Frederick Mosteller,et al.
Understanding robust and exploratory data analysis
,
1983
.
[3]
Glenford J. Myers,et al.
Composite/structured design
,
1978
.
[4]
Dennis G. Kafura,et al.
A validation of software metrics using many metrics and two resources
,
1985,
ICSE '85.
[5]
Sallie M. Henry,et al.
Software Structure Metrics Based on Information Flow
,
1981,
IEEE Transactions on Software Engineering.
[6]
Barbara A. Kitchenham,et al.
An evaluation of some design metrics
,
1990,
Softw. Eng. J..