<![CDATA[Software defects are a major issue in software development because they can affect the quality, reliability, and performance of a system. As digital technology advances at an increasingly rapid pace, software complexity is also rising, thereby increasing the likelihood of software defects. This study aims to apply and compare several machine learning and artificial intelligence methods for detecting software defects. The methods used in this study include Support Vector Machine (SVM), Principal Component Analysis (PCA) as a dimension reduction technique, and Backpropagation as a neural network-based method. The research process was conducted through a series of experiments to evaluate and compare the performance of each method based on the accuracy values obtained. The results show that the combination of SVM and PCA provides the best performance in detecting software defects compared to other methods. The highest accuracy obtained was 85.78% when using 13, 15, and 16 PCA components. Meanwhile, SVM without PCA achieved an accuracy of 85.47%, and Backpropagation achieved an accuracy of 84.83%. These results indicate that the application of PCA is capable of improving SVM classification performance through a dimension reduction process that preserves important features in the dataset. However, the performance achieved is still influenced by the characteristics of the dataset, the data distribution, and the model configuration used.]]>
Copyrights © 2026
