Meera Sharma

Predicting the priority of a reported bug using machine learning techniques and cross project validation

In bug repositories, we receive a large number of bug reports on daily basis. Managing such a large repository is a challenging job. Priority of a bug tells that how important and urgent it is for us to fix. Priority of a bug can be classified into 5 levels from PI to P5 where PI is the highest and P5 is the lowest priority. Correct prioritization of bugs helps in bug fix scheduling/assignment and resource allocation. Failure of this will result in delay of resolving important bugs. This requires a bug prediction system which can predict the priority of a newly reported bug. Cross project validation is also an important concern in empirical software engineering where we train classifier on one project and test it for prediction on other projects. In the available literature, we found very few papers for bug priority prediction and none of them dealt with cross project validation. In this paper, we have evaluated the performance of different machine learning techniques namely Support Vector Machine (SVM), Naive Bayes (NB), K-Nearest Neighbors (KNN) and Neural Network (NNet) in predicting the priority of the newly coming reports on the basis of different performance measures. We performed cross project validation for 76 cases of five data sets of open office and eclipse projects. The accuracy of different machine learning techniques in predicting the priority of a reported bug within and across project is found above 70% except Naive Bayes technique.

Understanding the meaning of bug attributes and prediction models

Software bug is a buzz word now a day. A software bug has many attributes, some of which are filled at the time of reporting and others are filled during the process of fixing. Some attributes are qualitative in nature but some are quantitative. A clear understanding of bug attributes, their interdependence and their contribution in predicting the other attributes will help in improving the quality of software. In the literature, prediction models based on linear regression have been proposed to predict the bug attributes and to determine their linear relationships. cc list (manpower involved in monitoring the progress of bug fix) is an important bug attribute for which no prediction model has been developed in literature. We investigated the contribution of bug attributes in predicting the bug cc list i.e. the manpower involved in monitoring the progress of bug fix based on multiple linear regression (MLR), support vector regression (SVR) and fuzzy linear regression (FLR). We conducted the experiments to develop prediction models for 21,424 bug reports of Firefox, Thunderbird, Seamonkey, Boot2Gecko, Add-on SDK, Bugzilla, Webtools and addons.mozilla.org products of the Mozilla open source project. We have also investigated a linear relation among bug attributes. The empirical results conclude that the value of R2 in predicting cc list across all datasets lies in the range of 0.31 to 0.70, 0.54 to 0.88, 0.25 to 0.68 and 0.69 to 0.93 for multiple linear regression, support vector regression, fuzzy linear regression(robust off) and fuzzy linear regression (robust bisquare) respectively.

Bug Assignee Prediction Using Association Rule Mining

In open source software development we have bug repository to which both developers and users can report bugs. Bug triage, deciding what to do with an incoming bug report, takes a large amount of developer resources and time. All newly coming bug reports must be triaged to determine whether the report is correct and requires attention and if it is, which potentially experienced developer/fixer will be assigned the responsibility of resolving the bug report. In this paper, we propose to apply association mining to assist in bug triage by using Apriori algorithm to predict the developer that should work on the bug based on the bugs severity, priority and summary terms. We demonstrate our approach on collection of 1,695 bug reports of Thunderbird, AddOnSDK and Bugzilla products of Mozilla open source project. We have analyzed the association rules for top five assignee of the three products. Association rules can support the managers to improve its process during development and save time and resources.

Prediction of the Complexity of Code Changes Based on Number of Open bugs, New Feature and Feature Improvement

During the last decade, a paradigm shift has been taken place in the software development process. Advancement in the internet technology has eased the software development under distributed environment irrespective of geographical locations. Result of this, Open Source Software systems which serve as key components of critical infrastructures in the society are still ever-expanding now. Open source software is evolved through an active participation of the users in terms of reporting of bugs, request for new features and feature improvements. These active users distributed across different geographical locations and are working towards the evolution of open source software. The code-changes due to bug fixes, new features and feature improvements for a given time period are used to predict the possible code changes in the software over a long run (potential complexity of code changes). It is evident that the open source software are evolved through these modification but an empirical understanding among the bug fix, new features, feature improvements and modifications in the files are unexplored till now. In this paper, we have predicted the potential of bugs that can be detected/fixed and new features, improvements that can be diffused in the software over a period of time. We have quantified the complexity of code changes (entropy) and after that predicted the complexity of code changes by applying Cobb-Douglas and extended Cobb-Douglas (two dimensions and three dimensions) based diffusion models. The developed models can be used to determine the quantitative value of complexity of code changes for reported bugs, new features and feature improvements in addition to their potential values. This empirical study mathematically models the interaction of a system (the debugging and code change system) with the external open world which will assist support managers in software maintenance activities and software evolution.

Clustering-based association rule mining for bug assignee prediction

Bug assignment is a decisive part of software maintenance. In this paper, we have proposed two approaches to apply association rule mining to assist bug triaging process. In the first approach, we have used apriori algorithm to predict the assignee of a newly reported bug based on the bugs severity, priority and summary terms. In the second approach, we have used X-means clustering followed by association rule mining inside each cluster. The redundant or identical meaning rules have been eliminated. We have analysed the association rules for top five assignees of Thunderbird, Add-on SDK and Bugzilla products of Mozilla open source project. We have also observed that the assignees who fixed Blocker and Critical bugs have less number of redundant rules in comparison of Normal bug fixers. Association rule mining after clustering results in rules with same or higher confidence.

Entropy Based Software Reliability Analysis of Multi-Version Open Source Software

The number of issues fixed in the current release of the software is one of the factors which decides the next release of the software. The source code files get changed during fixing of these issues. The uncertainty arises due to these changes is quantified using entropy based measures. We developed a Non-Homogeneous Poisson Process model for Open Source Software to understand the fixing of issues across releases. Based on this model, optimal release-updating using entropy and maximizing the active users satisfaction level subject to fixing of issues up to a desired level, is investigated as well. The proposed models have been validated on five products of the Apache open source project. The optimal release time estimated from the proposed model is close to the observed release time at different active users satisfaction levels. The proposed decision model can assist management to appropriately determine the optimal release-update time. The proposed entropy based model for issues estimation shows improvement in performance for 21 releases out of total 23 releases, when compared with well-known traditional software reliability growth models, namely GO model [1] and S-shaped model [2] . The proposed model is also found statistically significant.

Quantitative Quality Assessment of Open Source Software by Considering New Features and Feature Improvements

Open Source Software (OSS) evolves through active participation of users in terms of requesting for features, i.e. new features (NFs) and improvements in existing features (IMPs). Fixing of these features results in generation of further features improvements. In this paper, we have proposed a mathematical model to embody the OSS development based on the rate at which IMPs are generated as a result of fixing of features (NFs and IMPs). We have validated the model for datasets of five products, namely Avro, Pig, Hive, jUDDI and Whirr of Apache open source project. Results show that the model exhibit significant goodness of fit in terms of MSE (Mean Square Error), Bias, Variation, RMSPE (Root Mean Square Prediction Error) and R2 performance measures.

Complexity of the Code Changes and Issues Dependent Approach to Determine the Release Time of Software Product

Changes in source code of the software products are inevitable. We need to change the source code to fix the feature improvements, new features and bugs. Feature improvements, new features and bugs are collectively termed as issues. The changes in the source code of the software negatively impact its product, but necessary for the evolution of the software. The changes in source code are quantified using entropy based measure and it is called the complexity of code changes. In this paper, we built regression models to predict the next release time of software using the complexity of code changes (entropy), feature improvements, new feature implementation and bugs fixed. The regression models have been built using Multiple Linear Regression (MLR), various kernel functions based Support Vector Regression (SVR) and k-Nearest Neighbor (k-NN) methods. The proposed models have been empirically validated using four open source sub-projects of the Apache software foundation. The proposed models exhibit a good fit. The developed models will assist release managers in release planning of the software.

Developing Prediction Models to Assist Software Developers and Support Managers

A huge amount of historical information about the evolution of a software project is available in software repositories, namely bug repositories, source control repositories, archived communications, deployment logs, and code repositories. By mining the evolutionary history of a software, we have designed prediction models to assist software developers by predicting bug attributes like priority, severity, assignee and fix time. We have evaluated the uncertainty in the software in terms of entropy arises due to source code changes done in files of the software to fix different issues. To support software managers, we have designed prediction models to predict potential values of entropy and different issues, namely bugs, improvements in existing features (IMPs) and new features (NFs) over a long run. In this research work, we have developed mathematical models to assist software managers and developers in bug triaging, bug fixing and different software maintenance related tasks. Our work has been validated on issue and code change data of several open source projects, namely Eclipse, Open office, Mozilla and Apache.

Reduction of Redundant Rules in Association Rule Mining-Based Bug Assignment

Bug triaging is a process to decide what to do with newly coming bug reports. In this paper, we have mined association rules for the prediction of bug assignee of a newly reported bug using different bug attributes, namely, severity, priority, component and operating system. To deal with the problem of large data sets, we have taken subsets of data set by dividing the large data set using K-means clustering algorithm. We have used an Apriori algorithm in MATLAB to generate association rules. We have extracted the association rules for top 5 assignees in each cluster.The proposed method has been empirically validated on 14696 bug reports of Mozilla open source software project, namely, Seamonkey, Firefox and Bugzilla. The proposed method provides an improvement over the existing techniques for bug assignment problem.