| With the rapid development of society,people's pace of life is accelerating,and the highspeed pace of life can easily cause a series of physiological or psychological problems.Compared with adults with certain work experience and life experience,college students who have just left their family and friends are more likely to have problems,and even drop out of school or commit suicide.With the popularization and application of information technology,the data-driven anomaly detection technology has also achieved a series of excellent results in the field of student mental health education.Therefore,how to use anomaly detection technology to process and analyze student data on campus to assist universities to manage is a problem worthy of research.To solve this problem,this paper proposes a data-driven abnormal behavior detection and analysis system based on abnormal detection technology.The main work includes:(1)An improved algorithm combining Isolate Forest(iForest)and K-Means is proposed.The improved iForest algorithm mainly changes the branch construction process in the case of excessive branches and inconsistent scores existing in the traditional iForest algorithm.The method first uses the improved iForest algorithm to generate a set of anomaly candidates for the data set,and then uses K-Means to obtain the abnormal classification results from the anomaly candidates.(2)A student abnormal behavior detection and analysis system based on unsupervised anomaly detection algorithm is designed.The system not only has the function of student psychological assessment,but also can detect and analyze the abnormal behavior of students based on the improved iForest algorithm.(3)A student abnormal behavior detection and analysis system is realized.Based on jQuery and EasyUI front-end technology and SSH(Struts2+Spring+Hibernate)framework,this system takes MySQL as the database and Eclipse is used to develop behavior detection and analysis warning functions. |
