Predicting failures with developer networks and social network analysis

Abstract

Software fails and fixing it is expensive. Research in failure prediction has been highly successful at modeling software failures. Few models, however, consider the key cause of failures in software: people. Understanding the structure of developer collaboration could explain a lot about the reliability of the final product. We examine this collaboration structure with the developer network derived from code churn information that can predict failures at the file level. We conducted a case study involving a mature Nortel networking product of over three million lines of code. Failure prediction models were developed using test and post-release failure data from two releases, then validated against a subsequent release. One model's prioritization revealed 58% of the failures in 20% of the files compared with the optimal prioritization that would have found 61% in 20% of the files, indicating that a significant correlation exists between filebased developer network metrics and failures.

Social debt in software engineering: insights from industry

“…However, in our case we observed a live organisational and social structure to elicit possible causes and effects for its sub-optimality. While Nagappan et al establish the causality between organisational structures and software quality, we strived to understand patterns of sub-optimality across said structures, e.g., to allow for preventive action by means of social networks analysis (SNA) [15]. More in particular, we found correlations between sets of organisational-social circumstances and additional cost in software process.…”

An empirical study on predicting defect numbers

“…It has been widely recognized that the defect-proneness of software components (such as classes and code modules) is closely related to a considerable number of software metrics (the so-called features) [1], e.g., static code metrics, code change history, process metrics and network metrics [2], all of which are easy to collect now. Therefore, many defect prediction approaches using statistical methods or machine learning techniques have been proposed to forecast defect-prone software components [3].…”

Influence of confirmation biases of developers on software quality: an empirical study

“…Moreover, widely studied concepts such as attention, memory, motivation, personality, and social cognition are likely to affect software defect density. Interaction among software professionals during the software development process is also very likely to affect software defect density [40], [42], [43], [22]. Since, developers are the ones who implement and test code, static code metrics are reflections of human aspects, as well as other factors that are directly related to the software development process.…”

“…al. [40]. The model constructed for an industrial product from Nortel was able to explain 60% of the variance of failures during the testing phase.…”