Loading...
Please wait, while we are loading the content...
Similar Documents
Chapter 7 UML Metrics for Predicting Fault-prone Java Classes
| Content Provider | Semantic Scholar |
|---|---|
| Copyright Year | 2010 |
| Abstract | Identifying and fixing software problems before the implementation is believed to be much cheaper than during or after the implementation. Hence, it follows that predicting fault-proneness of software modules based on early software artifacts like software designs is beneficial as it allows software engineers to perform early predictions to anticipate and avoid faults early. Taking this motivation into consideration, in this chapter we evaluate the usefulness of UML design metrics to predict fault-proneness of Java classes. We use historical data of a significant industrial Java system to build and validate a UML-based prediction model. Based on the case study we have found that level of detail of messages and import coupling—both measured from sequence diagrams, are significant predictors of class fault-proneness. We also learn that the prediction model built exclusively using the UML design metrics demonstrates a better accuracy than the one built exclusively using code metrics. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://openaccess.leidenuniv.nl/bitstream/handle/1887/16070/07.pdf?sequence=10 |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Chapter |
