Aghabayli et al. describe a case study that integrates “software runtime data with development data in order to understand and predict” whether or not problems that occur “during the use of the software (external quality)” can be related back to problems that occurred “during the development of the software (internal quality).” The case study uses existing data from a Fraunhofer Institute for Experimental Software Engineering (IESE) project.
The authors use the cross-industry standard process for data mining (CRISP-DM) for the integration and analysis steps. The purpose of the analysis on the integrated data is to show to what extent development data can be used to identify software runtime quality problems. The results are inconclusive (weak correlation).
The challenge to future researchers is to make sure that the data collected during the software development and runtime processes can be used later for a specific research purpose. The authors conclude: “post-mortem retrofitting of the collected data to the actual problem to be solved might be very expensive or even impossible.” Therefore, even though the results obtained from the case study itself are not useful in and of themselves, the lessons learned can be useful to researchers.
As this topic should be of great interest to software development companies if the challenges can be overcome, they should assign at least one researcher to examining the subject rather than dismiss it out of hand. The section on related work suggests that, although there are some similarities, nothing that could be considered identical really exists. Readers should pay careful attention to this section, as well as open the links in the references, in order to get a better idea of whether they want to do their own research. But the minimum that any software developer should do is read the paper.