Research On Educational Data Mining Algorithms Based On Student Behavior Analysis

In the era of big data,the traditional student behavior management is increasingly exposed to the disadvantages of untimely intervention and the hysteresis of governance.Nowadays,with the application of educational big data in the analysis and monitoring of students' daily behaviors,managers can take the initiative to grasp the characteristics and rules of students' behaviors,and make research and judgment accordingly.With the development of information management systems(such as the student card system)in colleges and universities,it becomes easier and more convenient for us to collect and analyze students' behavior data.Extracting useful features from these behavior patterns plays a helpful role in understanding the student's learning process.Besides,students' behaviors are also important factors that can reflect students' learning styles and living habits on campus.At the same time,student's performance analysis and forecasting is a leading research direction of educational data mining.Many scientific research institutions apply machine learning technology to college projects through the open-source data obtained,which can not only help teachers understand the student's learning situation currently,but also benefit students' own staged development,and improve their whole performance in a personalized and targeted manner.In view of the existing studies,they mainly focus on extracting statistical features manually from the pre-stored data,resulting in hysteresis in predicting student's achievement and finding out student's problems.Furthermore,due to the limited representation capability of these manually extracted features,they can only understand the students' behaviors shallowly.In order to make the prediction process timely and automatically,this paper focuses on the construction of performance prediction model based on a variety of behavioral data generated by students in campus activities,which integrates data mining and deep learning techniques to treat the student performance prediction task as a short-term sequence prediction problem.The main research contents of this article are as follows:Firstly,the current research status of student performance prediction modeling and related data mining algorithms are summarized,and further introduces the specific algorithm ofsequential behavior modeling briefly.The precise classification and related applications of support vector machine algorithm in campus behavior data mining are analyzed.Secondly,in order to explore the effective value of the data in the campus card system of colleges and universities,this paper uses the data pre-processing technology to extract the relevant characteristics to analyze the campus card swiping behaviors,and applies mathematical statistics methods to study the factors that affect student achievement ranking.Thirdly,a feature modeling algorithm based on students' behavior sequences is proposed.On the basis of comprehensive consideration of students' behavioral intention and attention,this paper utilizes the sequence learning technology of hybrid recurrent neural network to realize accurate training of students' short-term behavioral characteristics,so as to improve the effect of classification and prediction.Finally,in the experiment,this paper takes the college student card system as an example to conduct empirical research,and verifies the effectiveness of the proposed hybrid method on real data.The experimental results clarify that comparing with the traditional student behavioral feature modeling methods,the behavioral modeling method based on deep learning can capture the students' intentions more accurately by combining the attention mechanism,so as to achieve the optimal classification effect.