• Corpus ID: 15316478

Defect Prediction in Software Projects-Using Genetic Algorithm based Fuzzy C-Means Clustering and Random Forest Classifier

@inproceedings{Suma2014DefectPI,
  title={Defect Prediction in Software Projects-Using Genetic Algorithm based Fuzzy C-Means Clustering and Random Forest Classifier},
  author={P. Suma and Ramaswamy V},
  year={2014}
}

• P. Suma, R. V
• Published 2014
• Computer Science

Software project success is based on prediction of defects at early stages of software development. Aaccurate prediction of defect prone modules in software development process enables effective discovery and identification of the defects. Such prediction approaches are valuable for the large scale systems, where verification experts need to focus their attention and resources to problem areas in the system under development. Identifying and locating defects in software projects to measure the project success is a difficult task. Especially, when project size grows, this task becomes expensive with sophisticated testing and evaluation mechanisms. Context: Software project failures are often caused by mistake made during the project, and such failures make a strong economic impact. It analysed software engineering projects such as product development project of a company, different software engineering projects and the projects were planned to be delivered to customers.Objective: This study reports an analysis for predicting defect prone modules using integrated approach of genetic algorithm based fuzzy c-means clustering with random forest algorithm.Method: This method was developed using Genetic Algorithm based Fuzzy C-means clustering with Random Forest classification applied on empirical data set and analysis was performed. Finally it is validated with five NASA public domain defect data sets. These data sets vary in size, but all typically contain a small number of defect samples in the learning set. For instance, in project PC1, only around 7% of the instances are defects.Results: The overall accuracy maximization is the goal, then learning from such data usually results in a biased classifier, i.e. the majority of samples would be classified into non-defect class.Conclusion: In order to achieve project success, it is better to understanding of the defects. It is reasonable to suggest that a notable portion of project defects are due to mistakes made during the project development in the case of accuracy based software engineering projects.