Study On Online Learning Behavior Analysis And Academic Warning Mechanism From The Perspective Of Self-regulation Learning

Relevant literature demonstrates the crucial importance of effectively utilizing self-regulated learning strategies for online learners.The ability of learners to achieve success largely depends on their capacity for online self-regulated learning.However,numerous challenges persist within the higher education system’s online learning environment,such as high dropout rates,subpar learning quality and efficiency,and the absence of personalized support.Therefore,the main research objectives include delving into the learning behavior patterns of online learners,enhancing learners’ self-regulated learning abilities,and providing personalized support.This paper focuses on a study conducted with 115 students enrolled in the "English Critical Reading 2" course on the Moodle platform at a university in S during one semester.Through empirical research,the study discusses the time series behavior patterns of different types of self-regulated learners and constructs an online academic warning model by incorporating meaningful variables of self-regulated learning behavior.The model encompasses the following three aspects:Literature review: This section defines the concepts of online self-regulated learning,data mining,academic early warning,and other related terms.It also reviews the current research status of online self-regulated learning and academic early warning both domestically and internationally.Based on Pintrich’s self-regulated learning theory model,the paper divides the analysis index of online self-regulated learning behavior,preprocesses the self-regulated learning behavior data from the Moodle platform,and identifies 14 significant self-regulated learning behavior variables closely associated with students’ academic performance.Data collection and analysis: Learners’ online self-regulated learning situations were collected through questionnaires.The K-clustering method was then employed to categorize self-regulated learners into three different types: high,medium,and low.By utilizing process mining technology,the study examined the time and sequential behavior patterns of these different types of learners.By analyzing and comparing the online learning behavior processes of learners with varying levels of self-regulated learning abilities,personalized learning support was provided,and suggestions were proposed for constructing an adaptive support system for the online learning environment and teachers’ instructional methods.Construction of an online academic early warning model: Eight different types of machine learning algorithms were utilized,including random forest,naive Bayes,and logistic regression,to construct an online academic early warning model.Through information gain rate analysis,self-regulated learning behavior indices with strong predictive abilities were identified.Recommendations were provided to help learners avoid academic risks.The results of the study indicate that,for learners in different grade levels(i.e.,freshmen,sophomores,and juniors),those with moderate to high self-regulated learning abilities exhibit greater significance in the retrospection task and planned retrospection events during the monitoring stage.They also demonstrate more consistency and relevance in the sequence and timing of resource viewing and discussion events in other self-regulated learning stages.Additionally,it was concluded that the naive Bayes model serves as the optimal prediction model for constructing academic early warning systems.The information gain rate analysis of the prediction indices revealed higher correlation coefficients and information gain values for variables such as the number of resource visits,the frequency of participation in discussions,and seeking help as self-regulated learning factors,indicating their greater contribution to the model.Consequently,teachers should pay closer attention to these indicators to prevent academic risks.In summary,this study contributes to the field by examining the behavior characteristics of online self-regulated learning and analyzing the online learning patterns of three distinct types of self-regulated learners(high,medium,and low).The findings validate that the self-regulated learning process occurs iteratively rather than nonlinearly,offering theoretical insights.Additionally,the study provides considerations for the design of adaptive online learning environments.Furthermore,the construction of an academic early warning model for online learning and the evaluation of academic.