An Efficient Software Defect Analysis Using Correlation-Based Oversampling

作者：Rao K Nitalaksheswara^*; Reddy Ch Satyananda

来源：ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING, 2018, 43(8): 4391-4411.

DOI：10.1007/s13369-018-3076-7

• Software defects analysis
• Classification
• Decision tree
• Class imbalance learning
• Oversampling

In software industry, probability of risks in large project is increasing due to the presence of defective modules which leads to failure in software execution. Quality assurance is one of the important aspects of software development. The extracted information from software repositories about defective modules can help project managers a lot for completion of projects on time with reliable quality. There are many existing knowledge extracting techniques for software defect analysis and prediction, but they have considered this problem in a general framework. Class imbalance is one of the major problems for constructing efficient decision trees for extracting knowledge from software defect datasets. To overcome this problem, a novel approach called as ICOS (Improved Correlation over Sampling) is proposed for handling class imbalance software defect datasets. This proposed approach uses oversampling strategy to generate new instances using synthetic and hybrid category approaches. The experimental results confirm that the proposed approach can efficiently identify the modules which are error-prone using simple rules.

• 出版日期2018-8