Research On Online Learning Behavior Data Analysis And Learning Effectiveness Prediction Method

With the development of the Internet era,educational informatization has been used more and popularized,and the development of online education and online courses has made educational informatization innovative.Since the launch of MOOCs in 2012,the forms of online learning have become more diversified,and the development of educational informatization has gradually become an important form of national education innovation and online education development.Prediction research on online learning effect of MOOCs is helpful to provide teaching AIDS for MOOCs learners,improve the efficiency of online learning and improve course content.Through the analysis of informatization,the research method combining entity connection and deep learning is constructed to analyze and study the relevant data of MOOC learning,which has certain research significance for the prediction of learning effectiveness of online education and the development of informatization teaching.This paper analyzes and studies the effectiveness of online learning on the data set of MOOPer,an education platform.MOOPer is an online open practice data set that collects online learning data of different users in MOOCs.By analyzing online learning information,users’ knowledge learning effectiveness can be analyzed,so as to build a corresponding analysis framework of online learning.By modeling users’ practice rating and learning content in the learning process,knowledge state and user learning effect structure are constructed to improve users’ knowledge mastery of online MOOC learning.In this study,the following work is carried out based on MOOPer data set:Firstly,the teaching situation data in MOOPer data set was preprocessed.The data information of MOOPer is relatively scattered,and the online learning data of users is also sparse.Some online MOOCs learners may not complete the complete learning of MOOCs due to various reasons during the learning process,so it is necessary to clean this part of data.In the process of data training,the model can better distinguish the more suitable learning styles for users at different stages,and at the same time,it can also establish the measurement standards of learning effectiveness for machine training.Secondly,on the basis of data preprocessing,the characteristic data of users’ online learning effectiveness are obtained by analyzing the number of users’ answers to the questions and the final score.The correlation of the data is obtained by correlation analysis.Meanwhile,the binary tree information selection in random forest is optimized and improved.The influence of random forest on data prediction was obtained by calculating the deviation point of the sample.The deviation distance of the machine learning model was calculated according to the real value and the predicted value.The improved random forest RF＿13 in this paper has improved the prediction performance of learning effectiveness compared with other machine models.Thirdly,correlation analysis was carried out according to the feature importance indicators in MOOPer data,and an improved model FLSTM of long and short term memory network was constructed.Matrix vector product operation was added to the improved model on the basis of the original,and the features extracted above were mapped into the output space of prediction data through nonlinear transformation,so as to improve the prediction of online learning effectiveness.Finally,the prediction accuracy of online learning effectiveness is compared between machine learning and deep learning,and the influence of different learning behavior characteristics of users on learning effectiveness in MOOC learning is analyzed.The comparison of knowledge tracking model to predict the effectiveness of online learning in data analysis can provide valuable exploration and thinking for further improving the evaluation of teaching prediction performance.