Indian Journal of Science and Technology

Open Access Open Access  Restricted Access Subscription Access

A Survey on Effective Bug Prioritizing using Instance Selection and Feature Selection Techniques


Affiliations
1 Department of Information Technology, Sathyabama University, Chennai – 600119, Tamil Nadu, India
2 Faculty of Computing, Sathyabama University, Chennai – 600119, Tamil Nadu, India
 

Background/Objectives: This paper represents a review of how to fix bugs and how to accurately assign a brand new bug to a particular developer. Software firms pay over 45 % of value in coping with computer code bugs. A complicated process of fixing bugs is bug triage that aims to accurately assign a developer to a brand new bug. To diminish the time value in human work, text classification techniques are used to conduct automatic bug sorting. Methods: Findings: In this paper, we possess a mind set to address the problem of knowledge reduction for bug sorting, i.e., the means to prune back the ratio and improve the standard of bug knowledge. We associate instance selection with feature selection to at the same time cut back knowledge ratio on the bug dimension and also the word dimension. To make away the parliamentary procedure of applying instance selection and feature election, we extract attributes from historical bug knowledge sets and build a prophetic model for a brand new bug knowledge set. We induce a tendency to thorough empirical observation and to investigate the execution of data reduction of the whole 600,000 bug reports of 2 giant open supply comes, namely Eclipse and Mozilla. Data reduction techniques such as classification and prediction method can be used to cut down the large bug dataset (by removing noisy and irrelevant data), and predict the correct answer for the bugs by applying keywords and properties. A classification method such as Bayesian Classification is used to classify the incoming bug dataset. Properties of Instance Selection and Feature Selection techniques are merged along with the attributes and keywords to solve the problem of bug prioritizing. Applications/Improvements: These outcomes demonstrate that the data reduction will effectively cut back the knowledge ratio and improve the accuracy of bug sorting. Our work provides associate approach to investment techniques on processing to create decreased and high-quality bug expertise in computer code development and sustenance. Regression analysis techniques can also be applied to a decision table to calculate the effort that is needed to resolve an incoming bug.

Keywords

Bayesian Classification, Data Reduction, Feature Selection, Instance Selection, Regression Analysis.
User

Abstract Views: 165

PDF Views: 0




  • A Survey on Effective Bug Prioritizing using Instance Selection and Feature Selection Techniques

Abstract Views: 165  |  PDF Views: 0

Authors

J. Sapphire Paul
Department of Information Technology, Sathyabama University, Chennai – 600119, Tamil Nadu, India
E. Brumancia
Faculty of Computing, Sathyabama University, Chennai – 600119, Tamil Nadu, India
S. Justin Samuel
Faculty of Computing, Sathyabama University, Chennai – 600119, Tamil Nadu, India

Abstract


Background/Objectives: This paper represents a review of how to fix bugs and how to accurately assign a brand new bug to a particular developer. Software firms pay over 45 % of value in coping with computer code bugs. A complicated process of fixing bugs is bug triage that aims to accurately assign a developer to a brand new bug. To diminish the time value in human work, text classification techniques are used to conduct automatic bug sorting. Methods: Findings: In this paper, we possess a mind set to address the problem of knowledge reduction for bug sorting, i.e., the means to prune back the ratio and improve the standard of bug knowledge. We associate instance selection with feature selection to at the same time cut back knowledge ratio on the bug dimension and also the word dimension. To make away the parliamentary procedure of applying instance selection and feature election, we extract attributes from historical bug knowledge sets and build a prophetic model for a brand new bug knowledge set. We induce a tendency to thorough empirical observation and to investigate the execution of data reduction of the whole 600,000 bug reports of 2 giant open supply comes, namely Eclipse and Mozilla. Data reduction techniques such as classification and prediction method can be used to cut down the large bug dataset (by removing noisy and irrelevant data), and predict the correct answer for the bugs by applying keywords and properties. A classification method such as Bayesian Classification is used to classify the incoming bug dataset. Properties of Instance Selection and Feature Selection techniques are merged along with the attributes and keywords to solve the problem of bug prioritizing. Applications/Improvements: These outcomes demonstrate that the data reduction will effectively cut back the knowledge ratio and improve the accuracy of bug sorting. Our work provides associate approach to investment techniques on processing to create decreased and high-quality bug expertise in computer code development and sustenance. Regression analysis techniques can also be applied to a decision table to calculate the effort that is needed to resolve an incoming bug.

Keywords


Bayesian Classification, Data Reduction, Feature Selection, Instance Selection, Regression Analysis.



DOI: https://doi.org/10.17485/ijst%2F2016%2Fv9i31%2F130972