Research On Modeling And Prediction Of Learning Motivation Based On Students’ Online Behavior And Performance

The continuous development of computer technology and Internet technology in the field of education has been paid more and more attention by educators and relevant personnel.These platforms have been concerned by all parties since their birth because of their breakthrough in the boundaries of time and space and the integration of high-quality resources.Especially in the context of COVID-19’s raging in 2020,online learning is more widely known and used by the public.However,compared with the traditional teaching form,online education platform which maintains complex information often produces massive information fragments due to the lack of proper maintenance.In the absence of effective guidance,students’ learning often stays at a shallow level,which leads to the learning effect is difficult to be guaranteed.In order to obtain effective learning results and ensure the service quality of online education platform,personalized learning guidance is needed to provide learners with more in-depth learning ways.Deep learning depends on providing learners with a more convenient and intuitive user experience and evaluating and strengthening their learning motivation.Therefore,in-depth understanding of the relationship from learning motivation to learning behavior to the final performance is very useful for the design of learning system architecture and the curriculum structure provided by the system.One of the main shortcomings of the existing research is the lack of explanation for the relationship between motivation and students’ participation behavior.Most studies use quantitative and qualitative methods to measure learning motivation through the analysis of data such as transcripts,interviews and questionnaires.The evaluation perspective of learners’ learning motivation is narrow,and the online behavior of learners in the online learning environment is not considered.and most of the existing research is based on the analysis of students’ performance in the whole semester of the course,and the final research results can not be timely fed back to the course at that time.Therefore,how to effectively analyze massive data,discover the characteristics of data hiding and make intelligent intervention to students with different learning motivation is the focus of the current "Internet plus" era of personalized teaching.To solve the above problems,this paper uses deep clustering algorithm to model learners’ behavior based on click stream data to analyze and predict learning motivation,the main contents are as follows:(1)Modeling and analysis of learning motivation dataBased on self-determination theory(SDT),this paper analyzes and models different learning motivations.This paper compares the participation characteristics of different learning motivation clusters,and uses Poisson regression and analysis of variance to analyze the correlation between each grouping variable.This paper studies the differences between students with different learning motivation and the potential relationship between learning motivation and education level.This paper analyzes the relationship between different groups of students inside and outside the school and demographic information such as age distribution.The experimental results show that students with external motivation will have better performance in the curriculum.The intensity of extrinsic motivation is positively correlated with students’ academic performance.The results verify the effectiveness of the proposed clustering model.(2)Construction and analysis of deep clustering modelIn order to analyze the learning behavior according to the click stream data of online learning platform to study different learning motivation.Based on the demographic statistics and click stream data about learning behavior recorded in the virtual learning environment of Open University,this paper proposes a deep clustering algorithm which combines Gaussian mixture model and stack autoencoder.The optimal number of clusters is determined by the contour coefficient and the error variance within clusters.The deep clustering model proposed in this paper is compared with the clustering model in other papers,and a comparative study is carried out from multiple perspectives.The results show that the model greatly improves the accuracy of clustering results through deep analysis of click stream data,makes the model more accurate to analyze the learning motivation of different groups of students,and makes up for the shortcomings of other clustering algorithms.The learning behavior analysis and motivation prediction model based on click stream data proposed in this paper can effectively judge the learning status of students,thus providing personalized reference for teaching activity strategies in online learning platform.This will help to improve the quality of teaching,improve the whole classroom environment,so that students can better enjoy the whole learning process.