【摘要】 传统学情分析主要依托作业与考试数据进行，存在数据滞后性强、反馈周期长等局限，而通过学情数据构建学习效果预测模型正是智慧教学体系建设的核心环节。以《电机学》课程为实践载体，依托超星学习通平台采集8个维度的学习行为数据指标，并融合先修课程成绩构建预测指标体系。通过对比六种机器学习模型的预测效能，结合Pearson相关系数探究学习行为与学业表现的相关性。实验结果表明，支持向量机(SVM)模型在准确率、召回率及F1-score三项指标上均达到接近90%的水平，显著优于其他对比模型。这不仅验证了本研究构建的指标体系的有效性，更证实基于SVM模型的预测方法具有较高的准确度，为智慧课程建设提供可靠的技术支持。更多还原

【Abstract】 Traditional learning situation analysis mainly relies on homework and examination data,which has limitations such as strong data lag and long feedback cycles.Constructing a learning effect prediction model based on learning situation data is a core part of the construction of an intelligent teaching system.Taking the course “Electromechanics” as a practical carrier,this study collects learning behavior data indicators of 8 dimensions based on the Chaoxing Learning Platform,and integrates the grades of prerequisite courses to construct a prediction indicator system.By comparing the predictive efficacy of six machine learning models and combining the Pearson correlation coefficient,this paper explores the correlation between learning behavior and academic performance.The experimental results show that the Support Vector Machine (SVM) model achieves nearly 90% in accuracy,recall rate and F1-score,which is significantly superior to other comparative models.This not only verifies the effectiveness of the indicator system constructed in this study,but also confirms that the prediction method based on the SVM model has high accuracy,providing reliable technical support for the construction of intelligent courses.更多还原