Yazılım hata tahmini için makine öğrenimi tasarlarmak ve uygulamak

The fundamental goal of software defect prediction (SDP) techniques (DPTs) is to find errors that hamper software development value (SDV). DPTs employ software history data to construct models, which are then used to anticipate faults in fresh occurrences of code areas such files, modifications, and procedures. The attempts made by earlier studies towards building accurate prediction frameworks are generally divided into the subsequent two categories: the primary tactic is physically designing original features or new amalgamations of features to characterize defects more efficiently, and also the second approach includes the utilization of nascent and better-quality classification frameworks supported machine learning. In the existing research work, genetic algorithm is applied with PSO for the feature extraction. The bagging classification is used to come up with final classified results. During this research work, three ensemble classifiers are generated to better understand the topic. When these ensemble classifiers are used for the PCA algorithm for the feature reduction, accuracy is increased further. In the last phase, the category(class) unbalancing problem gets resolved, and therefore accuracy is increased up to 93 percent. KEYWORDS: Accuracy, Amalgamations of features, Ensemble classifiers, Machine Learning, Prediction frameworks, Software Defect Prediction, Software Defect Prediction Techniques