Abstarct: In this research, using various software metrics including LOC, NOC, and WMC, as well as machine learning algorithms such as Decision Tree, Random Forest, and Multilxayer Perceptron Neural Network (MLP), software error prediction in Java classes has been investigated. A new model has been proposed for software error prediction in Java classes using a combination of these algorithms. In the proposed model, various features of Java source code are given as inputs to the model, and the output of the model is software error prediction in Java classes. To evaluate the accuracy of the proposed model, various metrics such as accuracy, precision, recall, and F1-score have been used. The results show that the proposed model has a high accuracy in predicting software errors in Java classes, and this model can be used as an advanced method for improving the quality of Java software. According to the results obtained in this research, the use of machine learning algorithms and neural networks in software error prediction in Java classes can have a significant improvement in the quality of Java software. Additionally, the proposed model in this research is suggested as an advanced method for software error prediction.