Analysis Of Learner's Behavior And Prediction Of Learning Risk Based On Statistical Learning Method

In today's information age,learning analytics is one of the emphases for research in the Educational Technology field.The realization of education informatization necessarily requires technical support.Learning analytics can connect technology and education.It is an effective way to promote further integration of technology and education.Above all,determine the purpose and object of the study and collect student learning data.Then,select data analytics methods and tools to analyze and model for student learning.Finally,feedback the results of the analysis to the teaching career to provide guidance for student learning.This paper start from learning analytics currently existing problems of Learning analytics model design.Applies statistical learning methods to learning analytics and conducting pilot study.This paper keeps track of the learning process of the students' four courses in the past two years of computer science and technology.There are two ways to collect student's learning data.The first way is to obtain student's learning behavior data from the online learning platform.The second way is use the real-name questionnaire survey to collect student offline learning data.Completes the following work:1.Investigate the basic learning situation of learners.Using the real-name questionnaire survey statistical analysis the basic learning situation of learners.And correlate these data with student's final exam score.Through comprehensive analysis,it is found that the main factors that have strong relevance to student's final exam score are the average score of student's pre-course,learning attitude,academic motivation,etc.2.Learner learning behaviors analysis based on Lasso-LARs method.Through the analysis of learner's behavior,the importance of each factor affecting student's final exam score is determined.The research find that academic motivation,average scores of pre-course,scores in experimental examination,exercise scores,preview time have the greatest impact on student's final exam score.After the learners and teachers can clarify the influence of these learning behaviors on learning outcomes,they can improve teaching and learning accordingly.3.Establishment of a learner classification model based on Support Vector Machine(SVM).The SVM algorithm is used to establish the student classification model.This model classifies students based on their learning behavior.4.Learner learning risk prediction based on Support Vector Regression(SVR).The SVR algorithm is used to establish a predictive model for student's final exam score.The model can predict student's final exam score based on student's learning behaviors and student's basiccharacteristics.Based on the correlation analysis,artificially change some learning behavior data of students,and verify the effectiveness of the intervention experimentally.This paper applies statistical learning methods to learner behavior analysis,learner classification modeling and learning risk prediction.And doing interdisciplinary pilot research.The research still have some shortcomings.In subsequent studies can further enrich the use of various methods and techniques in the Learning Analytics field.