摘要
传统的成绩预测方法没有严格区分学生学习行为的时序性和非时序性,因此提出学习行为分类融合的方法,首先根据学习行为特点划分为周学习特征和学期学习特征两大类,分别构建LSTM网络和MLP网络,两个网络融合后成为新模型的输入,最终实现学生学习成绩预测。将融合方法和传统预测方法进行对比,实验结果表明融合模型在预测准确度和F1值更高。
The traditional performance prediction methods do not strictly distinguish the timing and non-timing of students'learning behavior.Therefore,a classification and fusion method of learning behavior is proposed.Firstly,according to the characteristics of learning behavior,it is divided into weekly learning characteristics and semester learning characteristics,and LSTM network and MLP network are constructed respectively.After the fusion of the two networks,it becomes the input of the new model,and finally realizes the prediction of students'learning performance.The experimental results show that the fusion model has higher prediction accuracy and F1 value.
作者
李菲
曹阳
顾问
LI Fei;CAO Yang;GUWen(Sanjiang University,Nanjing 210012,Jiangsu)
出处
《电脑与电信》
2022年第8期27-31,共5页
Computer & Telecommunication
基金
2020年度江苏省高校哲学社会科学研究一般项目“在线开放课程中学习行为评估方法的研究”,项目编号:2020SJA0562。
关键词
在线学习
成绩预测
LSTM
网络融合
online learning
achievement prediction
LSTM
network concatenate
