Research On Data Mining Based On Learning Behavior In The Internet Environment

With the rapid development of Internet technology,the teaching mode has changed greatly,and the traditional learning mode has changed into the network learning mode gradually.The traditional teaching mode is the face-to-face teaching mode of the teachers and students,and the learning effect and learning status of the courses can be evaluated by on-site questioning and other methods.Online learning is a way of learning across time and space,and its learning effect and learning state method can not be evaluated by traditional methods,so how to evaluate online learning effect and learning status is very important.Learners in the process of learning to produce a large number of learning behavior data and be recorded,through the study of these data contained in the law,for the online learning platform for different learners to provide personalized environment and learning guidance,analysis and study of learner behavior data is very valuable.Based on the summarization and analysis of learning behavior,this paper uses method of classification data mining,proposes an integrated learning method to construct the Integrated Learning Achievement Classification Prediction model,and compares the performance of a single classification model and a stochastic forest integrated classification model constructed by decision tree algorithm and support vector machine algorithm.Through the construction of the performance prediction model,the learners' achievement prediction has been successfully realized.The main research work in this paper is as follows:(1)Based on the learning behavior data of platform learners,this paper analyzes and explores the basic information of learners ' learning behavior data,the types of learners and the factors influencing the achievement,and draws some conclusions Asingle factor variance(ANOVA)was used to analyze whether there were significant differences in learning behavior between different types of learners.For example,learners with different academic level have significant difference(P<0.05)in active days and forum posts two learning behaviors,but there are no significant differences in the number of learning events,the number of study chapters,and the proportion of modules completed in four aspects(p>0.05);There were significant differences among learners of different age groups in active days,forum postings,study chapters and module completion rations(p<0.05),while there was no significant difference(p>0.05)in the number of learning events.In addition,the five learning behavior characteristics of learners' learning events,active days,study chapters,forum posts and module completion ratios have different degrees of positive correlation(P<0.01).(2)In this paper,a single classifier is constructed by using the decision tree and support vector machine algorithm in the classification method,and the integrated classifier of stochastic forest construction is used to predict learners' academic achievement.Among them,the accuracy of the learning achievement Prediction model constructed by the decision tree is 0.812,the accuracy of the predictive model of SVM is 0.804,and the prediction model of the learning achievement based on the stochastic Forest integration classifier is 0.854.In the accuracy of the learning achievement Prediction model,the model of stochastic forest integration classifier is the best classification model,although the precision of decision tree algorithm and support vector machine algorithm is low,but the accuracy of the two is also more than0.8.Through the analysis of the experimental results,we can find out the relationship between learning behavior and academic achievement,so that the learning behavior and the combination of learning behavior will be more affected by the study.The educators of the platform can deeply understand the learner's autonomous learning and learning characteristics,thus better guide the teaching reform,optimize the teaching plan,and provide better service for the students.For the learners,can fully understand their learning behavior,improve learning efficiency,form a more effective way of learning.