Change Bursts as Defect Predictors

Nagappan, Nachiappan; Zeller, Andreas; Zimmermann, Thomas; Herzig, Kim; Murphy, Brendan

doi:10.1109/issre.2010.25

Cited by 153 publications

(110 citation statements)

References 19 publications

Supporting

Mentioning

106

Contrasting

Order By: Relevance

“…We also use the probabilistic classifier with Pre and NBD_max using the machine learning models for evaluating prediction efficiency. Probabilistic classifiers [19] assign a score or a probability to each sample. A probabilistic classifier is a function f: X -> [0, 1] that maps each sample x to a real number f(x).…”

Section: Experiments Results Analysismentioning

confidence: 99%

“…These metrics that are computed for classes or methods are aggregated by using average (avg), maximum (max), and accumulation (sum) at the file and package level. Another type of features is Change Metrics which is described in [19] In this paper, three versions of data which were collected weekly for the package level are integrated into one data file. All the complexity metrics and change metrics are also integrated.…”

Section: Eclipse Bug Datasetsmentioning

confidence: 99%

See 1 more Smart Citation

Using Source Code and Process Metrics for Defect Prediction - A Case Study of Three Algorithms and Dimensionality Reduction

Han¹,

Lung²,

Ajila³

2016

JSW

View full text Add to dashboard Cite

Software defect prediction is very important in helping the software development team allocate test resource efficiently and better understand the root cause of defects. Furthermore, it can help find the reason why a project is failure-prone. This paper applies binary classification in predicting if a software component has a bug by using three widely used algorithms in machine learning: Random Forest (RF), Neural Networks (NN), and Support Vector Machine (SVM). The paper investigates the applications of these algorithms to the challenging issue of predicting defects in software components. Thus, this paper combines source code metrics and process metrics as indicators for the Eclipse environment using the aforementioned three algorithms for a sample of weekly Eclipse features. In addition, this paper deals with the complex issue of data dimension and our results confirm the predictive capabilities of using data dimension reduction techniques such as Variable Importance (VI) and PCA. In our case the results of using only two features (NBD_max and Pre-defects) are comparable to the results of using 61 features. Furthermore, we evaluates the performance of the three algorithms vis-à-vis the data and both Neural Network and Random Forest turned out to have the best fit.

show abstract

Section: Experiments Results Analysismentioning

confidence: 99%

Section: Eclipse Bug Datasetsmentioning

confidence: 99%

Using Source Code and Process Metrics for Defect Prediction - A Case Study of Three Algorithms and Dimensionality Reduction

Han¹,

Lung²,

Ajila³

2016

JSW

View full text Add to dashboard Cite

show abstract

“…In the last decade, researchers have proposed a wide range of bug prediction models based on diverse information, such as source code metrics [3,27,48,47,32,46], historical data (e.g., number of changes, code churn, previous defects) [19,34,31,23,17,16], and developers interaction information (e.g., contribution structure) [37,40,25]. Since most prediction models were evaluated on different systems-and frequently with different performance measures-researchers have also investigated which approaches provide the best and most stable performance across different systems [24,30,41,9].…”

Section: Introductionmentioning

confidence: 99%

Method-level bug prediction

Giger

D’Ambros

Pinzger

et al. 2012

Proceedings of the ACM-IEEE International Symposium on Empirical Software Engineering and Measurement

134

172

View full text Add to dashboard Cite

Researchers proposed a wide range of approaches to build effective bug prediction models that take into account multiple aspects of the software development process. Such models achieved good prediction performance, guiding developers towards those parts of their system where a large share of bugs can be expected. However, most of those approaches predict bugs on file-level. This often leaves developers with a considerable amount of effort to examine all methods of a file until a bug is located. This particular problem is reinforced by the fact that large files are typically predicted as the most bugprone. In this paper, we present bug prediction models at the level of individual methods rather than at file-level. This increases the granularity of the prediction and thus reduces manual inspection efforts for developers. The models are based on change metrics and source code metrics that are typically used in bug prediction. Our experiments-performed on 21 Java open-source (sub-)systems-show that our prediction models reach a precision and recall of 84% and 88%, respectively. Furthermore, the results indicate that change metrics significantly outperform source code metrics. Method-Level Bug Prediction ABSTRACTResearchers proposed a wide range of approaches to build effective bug prediction models that take into account multiple aspects of the software development process. Such models achieved good prediction performance, guiding developers towards those parts of their system where a large share of bugs can be expected. However, most of those approaches predict bugs on file-level. This often leaves developers with a considerable amount of effort to examine all methods of a file until a bug is located. This particular problem is reinforced by the fact that large files are typically predicted as the most bug-prone. In this paper, we present bug prediction models at the level of individual methods rather than at file-level. This increases the granularity of the prediction and thus reduces manual inspection efforts for developers. The models are based on change metrics and source code metrics that are typically used in bug prediction. Our experiments-performed on 21 Java open-source (sub-)systems-show that our prediction models reach a precision and recall of 84% and 88%, respectively. Furthermore, the results indicate that change metrics significantly outperform source code metrics.

show abstract

“…Burnett [106] reflects on software quality issues in the context of end-user software engineering. The concepts behind change bursts [107], which allow to predict defects based on sequences of contributions, might be applicable to artifacts across layers. Concepts such as code ownership [52], and intellectual authorship in general could be incorporated.…”

Section: Discussionmentioning

confidence: 99%

Multilayered analysis of co-development of business information systems

Aram

Neumann

2015

J Internet Serv Appl

View full text Add to dashboard Cite

Business information systems (BIS) comprise technological (e.g. programs), informational (e.g. content) and social artifacts (e.g. collaboration structures). Typically, such systems are constantly and collectively developed (co-developed) further by a variety of individuals within the organization. By recognizing these varying types of actors (concerning their goals, technical expertise and language means) and their predominantly developed artifact type, one can distinguish two types of subsystems: technical subsystems wherein the development of the system behavior is conducted by software developers; and business subsystems dominated by end-users developing informational artifacts. So far, co-development structures within and between these subsystems are not well understood, especially the aspect that -potentially driven by appropriate measures such as the provision of domain-specific languagesco-development might shift between these subsystems. This paper presents an approach for characterizing the co-development of real-world BIS with respect to direct participation from different kinds of contributors. This multilayered approach allows us to analyze the co-development with programming languages, domain-specific languages and end-user tools. The approach is suited to assess the direct participation of individuals from different subsystems in the development of evolving BIS. We focus on the intersection of these subsystems, present appropriate metrics and a multilayered analysis scheme. Contributions to artifacts are analyzed using social network analysis to detect structural properties of continuous co-development. The application to Learn@WU, a real-world BIS, demonstrates how end-user enabling technologies have shifted the co-development effort of the system from a small group of developers to a several orders of magnitude larger group of contributors. We observed an increase of direct participation over time on both informational and executable artifacts, while the number of technical experts was more or less constant. Our approach may act as a trigger for the application and further development of rigorous instruments for assessing co-development of BIS.

show abstract

Change Bursts as Defect Predictors

Cited by 153 publications

References 19 publications

Using Source Code and Process Metrics for Defect Prediction - A Case Study of Three Algorithms and Dimensionality Reduction

Using Source Code and Process Metrics for Defect Prediction - A Case Study of Three Algorithms and Dimensionality Reduction

Method-level bug prediction

Multilayered analysis of co-development of business information systems

Contact Info

Product

Resources

About