基于数据驱动的大学生心理健康风险识别研究  

Research on Mental Health Risk Identification of College Students Based on Data Drive

作　　者：孙聪 刘大旭[1] 王冠卓[1] SUN Cong;LIU Daxu;WANG Guanzhuo(Jiamusi School,Heilongjiang University of Chinese Medicine,Harbin 150040,China)

机构地区：[1]黑龙江中医药大学,佳木斯学院,黑龙江哈尔滨150040

出　　处：《微型电脑应用》2023年第8期72-75,共4页Microcomputer Applications

基　　金：四川医院管理和发展研究中心项目(SCYG2022-21)。

摘　　要：为了提高大学生心理健康风险识别的准确性和稳定性,提出基于数据驱动的大学生心理健康风险识别方法。利用改进的iForest算法从数据库中读取到的大学生心理健康数据中筛选出异常候选数据,采用基于凝聚k-means的决策簇分类器,从建造在由异常候选数据组成的训练数据集上的一系列自上而下的嵌套式聚类中的树上提取分类模型,并从中抽出一些含有高置信度的决策簇来分类未标记样本,实现大学生心理健康风险识别。实验结果说明,所提方法对不同刺激下以及不同专业大学生心理健康风险识别的精度均高于95%,稳定性强,且当簇数目为2时所提方法的分类区域特征可以更好地被表示出来。In order to improve the accuracy and stability of college students'mental health risk identification,a data-driven method for college students'mental health risk identification is proposed.The improved iForest algorithm is used to screen out the abnormal candidate data from the mental health data of college students by database.The decision cluster classifier based on agglomerated k-means is used to extract the classification model from a series of top-down nested clustering trees built on the training data set composed of abnormal candidate data,and extract some decisions with high confidence cluster,and is used to classify unlabeled samples to realize mental health risk identification of college students.The experimental results show that the accuracy of the proposed method for mental health risk identification of college students with different stimuli and different majors is higher than 95%,and the stability is strong.When the number of clusters is 2,the characteristics of the classification region of the proposed method can be better expressed.

关 键 词：数据驱动 心理健康 风险识别 决策簇 分类器 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]