基于机器学习的在线学习数据分析与研究

Analysis and Research of Online Learning Data Based on Machine Learning

本文主要以机器学习方法为基础对学生在线学习行为数据进行分析和研究。首先采用多元线性回归的方法对从网络教学平台获得的14个量化特征按照对考试成绩的影响权重进行排序,随后采用广义回归神经网络分别对选出的权重较高的9个特征以及全部特征进行建模。结果表明,通过特征选择后的模型平均绝对误差为5.7,无论是在预测能力还是拟合优度方面都要高于采用全部特征的模型,因此该模型可以有效预测考试成绩,使教师能够识别出容易不及格的学生,及时提供干预与支持。

This paper mainly analyzes and studies students'online learning behavior data based on machine learning methods.Firstly,the multiple linear regression method is used to rank the 14 quantitative features obtained from the online teaching platform according to the weights of the test scores,Then,nine features with higher weight and all features are modeled by using the generalized regression neural network.The results show that the average absolute error of the model after feature selection is 5.7,which is higher than the model with all features in both predictive ability and goodness of fit.Therefore,the model can effectively predict test scores and enable teachers to identify Students who are prone to fail,provide timely intervention and support.