The reliability of software depends on its ability to function without error. Unfortunately, errors can be generated during any phase of software development. In the field of software engineering, the prediction of software defects during the initial stages of development has therefore become a top priority. Scientific data are used to predict the software's future release. Study shows that machine learning and hybrid algorithms are change benchmarks in the prediction of defects. During the past two decades, various approaches to software defect prediction that rely on software metrics have been proposed. This paper explores and compares well-known supervised machine learning and hybrid ensemble classifiers in eight PROMISE datasets. The experimental results showed that AdaBoost support vector machines and bagging support vector machines were the best performing classifiers in Accuracy, AUC, recall and F-measure.