Software Defect Prediction System Based on Decision Tree Algorithm

Abstract

Software defect prediction plays a crucial role in ensuring software quality and minimizing the potential risks associated with defects. This study aims to develop a comprehensive software defect prediction system that utilizes tree-based algorithms to enhance accuracy, feature selection, and evaluation metrics. The study addresses the limitations of previous research by considering a broader range of datasets, comparing computational efficiency with other ensemble techniques, and examining the impact of hyperparameters on model performance. The implemented system consists of three stages: dataset loading, processing, and result presentation. The dataset loading page allows users to upload their datasets in CSV format, simplifying the prediction process. The processing page performs essential tasks such as feature engineering, normalization using minimax normalization, and training the model with the decision tree algorithm. These steps ensure the extraction of relevant features, transformation of data, and learning of patterns and correlations for accurate software defect prediction. The study emphasizes the practical implementation of the developed system, going beyond model evaluation. By providing a fully functional and integrated system, this study bridges the gap between research and real-world application. The findings of this study contribute to the field of software defect prediction by offering an improved system that enhances accuracy, feature selection, and evaluation metrics. This has implications for software development and quality assurance processes, ultimately leading to higher software quality and increased productivity.