Statistical Identification And Psychological Intervention Of High Risk Participants In MOOC

Massive Open Online Courses(MOOC)have broken the time-space limit of traditional teaching,with rich Online course content and wide selection of students.MOOC education has become a strong support for the development of emerging education.However,as MOOC is a self-learning model with no supervision and no consequences,it is difficult for MOOC participants to maintain their learning attention and learning continuity,resulting in low pass rate and high dropout rate.In-depth exploration of students' online learning behavior patterns,accurate identification of high-risk MOOC participants,and timely intervention of their learning behaviors with scientific methods have become effective ways to improve the passing rate of MOOC courses and improve the teaching effect.In this study,4144 pieces of participants' information were extracted from the course probability theory and mathematical statistics on the MOOC website "xuetaon.com"--including 90 indicators,such as the completion of participants' homework,online discussion,video learning duration,and tests.Prior to the imbalance in the completion of the sample indexes,use the statistical analysis tool of R language to treat the existing unbalanced data in a balanced way,using the SMOTE(Synthetic Minority Oversampling Technique),and obtain 279 uniformly classified samples.Then,statistical identification of high-risk participants was carried out for the data set of 279*7 combined with the integrated algorithm XGBoost.Through manual screening and Lasso regression variable selection,it was finally found that the indicators that affect students' final course exam scores include achievement completion degree,learning duration,video watching,discussion forum behavior and learning of relevant chapters.Kappa increases from 1.301819 to 282.7768,avoiding multicollinearity among variables.Under the ten-fold cross-validation,XGBoost algorithm was used to predict the results of 279 students' performance statistics,with an average prediction accuracy of 99.6%.Subsequently,the network learning related psychology intervention study,multimedia technology,peer feedback,cooperative learning,simulation games,even compulsory constraints on student's learning behavior,learners can effectively reduce the network soul roaming and low concentration,eventually improve the learning efficiency,to ensure the effective implementation of the network curriculum.This study shows that existing indicators and statistical methods can be used to make effective prediction analysis on whether MOOC participants finally pass.In the online students' learning process intervention,the lecturer's sense of participation is too low,so the instructor can join the online learners' learning process,guide students to approach the discussion area,and improve students' mutual feedback behavior.Teachers can also encourage students' cooperative behavior during lectures,providing a good and healthy learning atmosphere for participants in online learning.