Advanced EdTech Strategies for Predicting Software Reliability: Harnessing ALM and SVM

Software defect prediction is pivotal for ensuring the reliability of new software systems. This study introduces an innovative approach to the domain, leveraging software reliability prediction. This study is focused on solving two critical questions: the utility of historical failure data and the relevance of specific failure datatypes. Furthermore, this investigation examines the trade-off between prediction accuracy and timeliness in SVM and ELM-based models. Experimental results highlight the superior predictive power of ELM, offering enhanced metrics such as specificity, regression, accuracy, and F1 score. In addition to predictive models, this research presents a sophisticated feature selection framework utilizing ELM and SVM. The NASA Metrics dataset serves as a robust testing ground. To address dataset imbalances, resampling techniques are applied to optimize prediction accuracy prior to feature selection. In sum, this study offers a concise yet comprehensive approach to software reliability prediction, advancing the field's capabilities and contributing to elevated software quality assurance practices