Font Size: a A A

Hierarchical Algorithm Of College Students And Abnormal Students Detection Based On Behavior Data

Posted on:2021-05-17Degree:MasterType:Thesis
Country:ChinaCandidate:Z C LiFull Text:PDF
GTID:2427330605963437Subject:Applied Statistics
Abstract/Summary:PDF Full Text Request
The education of students has always been the top priority of colleges and universities in China.In the implementation of quality education today,how to carry out accurate education for students has become the primary issue.Under the limited conditions of teacher resources,human resources and management resources,how to use Internet technology to solve the problem of educational stratification is particularly important.In recent years,the extracurricular activities of college students are increasingly rich and campus life is more diverse.With the rise of mobile Internet and e-commerce,students can deal with teaching affairs and daily life needs without leaving home,which also makes the management of students difficult and complex.This paper selects the behavior data of some undergraduates in our university,including all-round data such as the length of participating in extracurricular activities,library related data,canteen data and scores,to make a level division of students,and train a management classification model based on students through machine learning.To realize the division of students,to facilitate counselors,teachers and other personnel to manage students.In this paper,first of all,the idea of clustering analysis is used to analyze the students' performance data and behavior data to carry out hidden rating,calibrate and audit the clustering results,adjust the algorithm and parameters to get the appropriate clustering results.Compared with the results of the introduction of behavioral data,the effectiveness and feasibility of behavioral data can be clearly found,which can be passed through the evaluation of counselors or teachers.Then we turn to the next stage of the classification model,training a suitable model to predict the new student data,outlier detection is an important part of learning situation detection,through the improved algorithm for the effective calibration of abnormal students,screening out suspicious students to report processing,and then realize the hierarchical and accurate education of students.
Keywords/Search Tags:precision education, big data, behavior data, machine learning, classification algorithm, confidence index, abnormal value analysis of learning situation detection
PDF Full Text Request
Related items