Enhancing Software Defect Prediction Using Principle Component Analysis and Self-Organizing Map

Application of SDLC in software is needed to produce a quality software. One of the way to improve software quality is determined by checking and testing, to find defects in the software that can degrade software quality. Software defect datasets generally have problems with class imbalance and unrelated features that may cause performance degradation in learning algorithm. This research proposes the selection of Principle Component Analysis (PCA) feature to solve the problem of unrelated features, while to overcome the problem of class imbalance using Self Organizing Maps model (SOM) by compare some data classifications algorithm to find the optimum result. The experimental results in this study shows that Random Forest get the highest average of accuracy 96.87%, average of precision 96.88%, and average of recall 96.88%. For JM1 dataset got the highest average of 96.06% accuracy, average of 96.44% Precision, and average of 96.06% Recall.