Recognition And Research Of Students With Abnormal Psychology Based On Educational Big Data

With the improvement of education popularization,the number of college students is increasing,and there are more and more students with psychological problems.Whether students with abnormal psychology can be found in time is one of the main problems facing colleges and universities.Big data of university education plays a vital role in analyzing and identifying students with abnormal psychology.This article is based on the educational data collected by the school card,educational administration system,access control system and related business systems,and extracts the behavioral characteristics of students through data cleaning and transformation of the massive and messy school data of students.And through hypothesis testing to analyze the differences in school behavior characteristics between normal students and students with abnormal psychology,and finally establish a model for identifying students with abnormal psychology and evaluate the results.The specific research content and innovation are as follows:(1)By collecting students' school data,preprocessing such as data cleaning,integration and data transformation,and extracting the characteristics of students' social relationships,regularity of life,diligence,academic performance and economic status at school.Finally,it compares and analyzes the behavior characteristics of abnormal students and normal students in school,and finds out the similarities and differences of the two types of students in school performance.(2)An adaptive threshold algorithm based on association rules is proposed to mine the social relationships of students in school.Based on the principle of co-occurrence of student card data in multiple locations,an adaptive method is used to determine the threshold according to the sample size of student swiping data and the number of co-occurrences.And use the real social network of some students to verify the threshold,and finally ensure the accuracy of the students' social relations in school.(3)On the basis of extracting behavioral characteristics,by processing extremely unbalanced data samples,establish a model for identifying students with abnormal psychology.The experimental results show that using the optimized logistic regression model to classify normal and abnormal students,the accuracy and G-mean value are 83.6%and 83.1%,respectively.In addition,through experiments,it is found that to identify students with psychological abnormalities,at least one year of school data of students is required.In summary,this article proposes a model for mining the social relationships of students at school based on educational big data,and extracts the behavioral characteristics of students at school,and establishes a model for identifying students with abnormal psychology.Experiments have proved that this model can effectively identify students with abnormal psychology and promptly allow student managers to intervene and deal with it.