<p>The LPSO phase structure is a key reinforcing phase in magnesium-rare earth (Mg-RE) alloys, and the appropriate morphology of the LPSO phase structure can significantly enhance the mechanical properties of the alloy. In this paper, we construct a dataset for Mg-RE alloys and set the presence or absence of LPSO phase structure in the alloys as the main goal of the classification task. Based on the constructed Mg-RE alloy dataset, five algorithms-k-nearest neighbor, support vector machine, random forest (RF), gradient boosted decision tree, and multilayer perceptron (MLP) were utilized to build classification prediction models, and all of these models achieved an accuracy of 0.9. Among these models, the RF model with the highest prediction accuracy was chosen as the main prediction tool, and a tree with the highest accuracy was further selected from its forest as the final classifier. By visualizing this decision tree, a suitable range of alloy compositions was determined. Subsequently, combining the advantages of RF and MLP algorithms, an integrated B-MLP classification prediction model was innovatively developed. On the test set, the B-MLP model shows excellent performance with an AUC value of 0.99 and outperforms any single classification model in all metrics such as accuracy, precision, recall, and F1-score. Compared to traditional comparative experimental methods, machine learning techniques are able to more rapidly and accurately dissect the multiple factors affecting the phase structure of LPSOs due to their ability to analyze big data. The range of alloy compositions derived from machine learning provides a crucial guideline for the preparation of Mg-RE alloys containing the LPSO phase structure, which greatly improves the efficiency and accuracy of material design and development.</p>