Study Of Learners' Behavioral Characteristics In The Context Of MOOC Fever

With the explosive development of Internet technology and the democratization and popularization of education,the product of "Internet+education" has emerged,that is,the large-scale open online course MOOC(massive open online course).Compared with the curriculum scale of students in higher education all over the word,the huge learning group of MOOC produces a large amount of learning behavior data,which can be described,counted,assessed and analyzed through relevant statistical analysis methods.This helps to understand the learning behavior of MOOC learners,optimize the MOOC platform and improve the teaching quality.The main research content of this paper is the classification of learner's behavior on MOOC platform.The sample is an open data set on learner's learning behavior provided by Canvas Network.This paper classifies the learner's learning behavior data by machine learning algorithms,and compares the accuracy of gradient boosting decision tree(GBDT),naive Bayesian,neural network and support vector machine in classification analysis.The results show that GBDT has highest accuracy for learner type.The learners' learning behavior was analyzed by two-step cluster analysis based on SPSS21.0.The results show that three is the optimal clustering number,which can make the distribution of samples more reasonable and the characteristics more obvious.This paper describes the types of learners after classification by combining both of the conclusions,which has practical reference for the study of learners'learning behavior.