Research On Personalized Teaching Intervention Of MOOC Courses Under The Perspective Of Big Data And Machine Learning

The rise and development of MOOC provides high-quality educational resources for learners around the world.While promoting education equity and breaking through learning time and space constraints,low completion rate has become an important factor hindering its development.In recent years,researchers in the education field have proposed a personalized teaching concept,but in the long-term development of online education,personalized teaching is still difficult to achieve a good implementation.The reason is mainly due to its large number of MOOC learners and the large proportion of invalid learners and resource viewers.It is difficult to locate and individualized teaching interventions for a small number of effective learners,which has become a major problem for MOOC.Based on edX open dataset,this paper constructs and validates the method flow of personalized instructional intervention for MOOC learners with data analysis,predictive model and simulation,and it also designs Prediction-Analysis-Intervention,an iterative personalized teaching intervention framework that accurately intervenes in learners who are unable to graduate,to help them complete the courses and obtain a certificate.Firstly,from the perspectives of learner achievement,learner characteristics,learning behavior,correlation analysis and multivariate analysis,the courses and learners in MOOC are deeply explored with descriptive data analysis and exploratory data analysis,to research the characteristics of the course learner,the distribution of learning behavior and performance,and the correlation between various data information.Through descriptive data analysis,it is found that there are a large number of invalid learners and resource viewers in MOOC,effective learners occupy a very small proportion,and each information of each course learner presents a very different distribution characteristics.Through exploratory data analysis,it is found that learners'data information and performance have different correlations,and there are also linear or nonlinear complex relationships between them.Therefore,all kinds of information have different degrees of influence on learning effects,and have important reference value for learners' personalization.Secondly,based on the data information in MOOC,the learner's scores are predicted by linear regression,multiple linear regression and deep neural network.It is found that multivariate linear regression comprehensively considers the factors of curriculum and learners based on multivariate,it can achieve more accurate predictions of learner performance than linear regression of single variables,and the deep neural networks,its regression can better fit the complex nonlinear relationship between various data and information,and achieve more accurate prediction of learner's performance than multivariate linear regression.Through the evaluation and comparison of all the models,the paper finally selected the deep neural network as the prediction model of the results.The model reduces the cost of evaluating the learning effect of a large number of learners,and eliminates the subjective factors of manual evaluation,and achieves accurate assessment of learners,so as to accurately predict the learners who may not be able to graduate among the mass learners.Thirdly,for those learners who may not be able to graduate,the learners who are closest to the information from the graduated learners are selected,and this graduated learners are used as reference objects to analyze the differences between them.The reason for the failure to graduate is explored,and a personalized teaching intervention for the learner is designed based on the results of the analysis.Through simulation experiments,the learners' behavioral performance after teaching intervention was simulated,and the learner's scores before and after the intervention were compared,to verify the effectiveness of the prediction and intervention.The experiment proves that the method is effective for the intervention of the learner,and achieves accurate positioning and improvement of learning effect,and improves the average score and graduation rate of the MOOC courses.Finally,based on the whole research content,design the personalized teaching intervention framework of Predictive-Analysis-Intervention,with performance prediction and early warning of learners who can't graduate,match the learners'reference objects to analyze the reasons why they can't graduate,and conduct personalized teaching intervention for them in a targeted manner,with iterative process,to help MOOC learners complete the course and graduate,Based on big data analysis,machine learning and personalized teaching theory,this paper explores and designs MOOC's personalized teaching plan.Through analysis and experiment,it gradually constructs and verifies the personalized teaching intervention framework of Predictive-Analysis-Intervention.The research shows that the method is effective in practical application scenarios,and effectively promote the implementation of personalized teaching in MOOC,so as to improve the experience and effect of learning.