Research On Online Learner Behaviour Mining And Classification Model Based On Deep Learning

In recent years,data-driven education reform,innovation and upgrading have become effective ways to improve higher education,but there are also limitations in terms of weak practical skills,lack of self-adaptability and inefficient learning.To ensure the effective implementation of online learning,more in-depth research must be conducted on the identification of online learner types,and more modern learning technologies must be applied to enrich the theories and applications in the field of online learning.Based on the existing research,this thesis constructs a classification model for identifying learner types based on the analysis of online learning behaviour from the learner’s perspective,and specifically makes the following three researches: firstly,we identify and filter the factors influencing the classification of online learners based on LFFA.Based on behavioural science theory and social cognitive theory,as well as behavioural components and attributes,we construct a behavioural indicator system that influences learner classification,and use the LFFA algorithm to filter features and eliminate irrelevant factors to obtain an indicator system containing five key behavioural features for subsequent modelling and prediction.The Mo E-LSTM model,LFFA-LSSVM and Deepforest model are used for experiments to improve the recognition of online learners’ types;Finally,online learner classification models were constructed and experimented with according to the Stacking architecture.The three recognition models from the previous section were used as primary learners,and the Stacking architecture was applied to fit the three models using the idea of integrated learning.The prediction results were output in the form of a progressive gradient tree,and the results were validated using a ten-fold crossover.After the experimental comparison and analysis,it is found that compared with the traditional single recognition model,the online learner classification and recognition model based on deep learning constructed in this thesis has better prediction accuracy,deepening the theoretical research and application of deep learning models in online education and behavioural recognition prediction,which can solve the problem of inaccuracy of previous research data,and can effectively identify and portray learners in order to tailor teaching to their needs and It can adjust teaching methods and make reasonable teaching plans,thus improving the level of online education.