基于时间序列数据驱动的在线学业预测机理模型研究  

Research on Mechanism Models for Online Academic Performance Prediction Using Time-Series Data

作　　者：姜强[1] 刘盼[1] 倪静 郝美霞 赵蔚[1] JIANG Qiang;LIU Pan;NI Jing;HAO Meixia;ZHAO Wei(Northeast Normal University,Changchun Jilin 130117;Qingdao Chengyang Experimental High School,Qingdao Shandong 266109)

机构地区：[1]东北师范大学,吉林长春130117 [2]青岛市城阳区实验高级中学,山东青岛266109

出　　处：《现代远距离教育》2025年第1期58-67,共10页Modern Distance Education

基　　金：国家自然科学基金面上项目“网络学习空间中的学习风险预警模型和干预机制研究”(编号:62077012);教育部哲学社会科学研究后期资助项目“高质量发展目标下深度学习的发生机制与促进策略研究”(编号:23JHQ086);吉林省社会科学基金项目“吉林省高等教育数字化转型路径及对策研究”(编号:2023B88);吉林省教育厅社科重点项目“数据驱动的在线学业预警及干预策略研究”(编号:JJKH20241387SK);吉林省高等教育教学改革研究课题“数智时代混合式教学模式下教育技术学专业课程建设与教学实践研究”(编号:2024L5L2IY6001U)。

摘　　要：在线学业预测是教育领域中的关键环节,是实现教育数字化和个性化教学的重要途径。当前,在线学业预测方法多依赖数据的静态特征,动态捕捉学生的在线学习行为是确保学业预测精准度和适用性的迫切要求。本研究以数据驱动和人工智能技术为核心,提出在线学业预测框架,以视频点击流时间序列数据为基础,聚焦在线学业预测关键特征,利用长短期记忆网络(LSTM)构建在线学业预测机理模型。此外,应用预测模型开展实践,验证在线学业预测模型效果,利用学习分析仪表盘可视化反馈结果,实现个性化学习和精细化教学。研究表明,基于时间序列数据的在线学业预测机理模型能够精准追踪学生的学业表现,并在实际应用中表现出优异的预测精度和稳定性。研究成果在推动教育数字化转型、深化个性化教学实践以及提升教育决策精准性等方面提供了重要的理论和实践依据。Online academic performance prediction is a critical aspect of the education field and an essential pathway for achieving educational digitalization and personalized teaching.Currently,online academic prediction methods mostly rely on the static characteristics of data,and realizing dynamic capture of students’online learning behaviors is an urgent requirement to ensure the accuracy and applicability of academic prediction.This study proposes an online academic performance prediction framework driven by data and artificial intelligence technologies.The framework uses video clickstream time-series data to identify key features of online academic performance and employs an LSTM(Long Short-Term Memory)network to construct a mechanism model for prediction.Additionally,the predictive model is applied in practice to validate its effectiveness.A learning analytics dashboard is used to provide visual feedback on results,enabling personalized learning and refined teaching strategies.The findings demonstrate that the time-series-based online academic performance prediction mechanism model can accurately track students’academic progress and exhibits prediction accuracy and stability in practical applications.This research provides significant theoretical and practical support for advancing educational digitalization,deepening personalized teaching practices,and enhancing the precision of educational decision-making.

关 键 词：时间序列数据 在线学习 学业预测 机理模型 

分 类 号：G43[文化科学—教育学]