基于微博的青少年心理压力趋势预测

发布时间：2018-02-23 19:03

本文关键词： 青少年压力微博预测时间序列 K线图　出处：《清华大学》2015年博士论文　论文类型：学位论文

【摘要】：青少年们已经成为社会中最有压力的群体之一,调查显示,各国的青少年均遭受严重的压力困扰,自杀更是成为青少年的重要死因。因此,及时了解青少年的压力状况,预测其可能的心理压力趋势,成为一个重要的研究问题。随着社交网络的快速发展,越来越多的青少年青睐于使用这种生动的富媒体交互方式来获取信息及自我表达,这为我们通过微博来获得青少年的心理压力信息成为了可能。本文通过微博来预测青少年的心理压力趋势,这些压力预测结果可以对青少年起到预警的作用,有压力的青少年可以及时调整他们的行为方式,采取合理的应对措施来缓解压力,以防压力的长期累积导致严重的心理问题。青少年的老师及监护者也可以更好地了解青少年的压力状况,提供及时的个性化指导,帮助青少年们释放压力。本文的主要研究工作包括以下几方面:1.预测青少年下一个时刻的心理压力值。本文通过对青少年微博中检测出的压力数据进行数据聚集、缺失值填补等数据预处理工作,得到了多因素的压力时间序列。论文采用了多因素时间序列模型来对青少年的压力变化规律进行建模,并进一步考虑事件对青少年心理压力的影响,通过事件趋势挖掘来得到事件压力变化模式,此对时间序列模型的预测数据进行修正,从而实现更加准确的预测。实验结果表明,通过考虑多个压力因素及事件修正,可以明显改善预测效果。2.预测青少年下一个时刻的心理压力值变化。基于青少年心理压力变化与股市价格波动的相似性,论文引入股票趋势分析中最常用的K线理论来表示青少年的压力值信息,并通过基于K线的压力模式匹配,从历史K线序列中捕捉压力趋势变化信号,实现了对下一个时刻的压力值变化趋势的预测,即压力增大、减小还是不变。实验表明基于K线的压力值变化趋势预测模型的预测误差为16.21%,优于其它几种趋势分析方法。3.预测青少年未来的心理压力变化幅度。论文通过K线理论,结合模糊的数据表示来预测青少年未来的心理压力变化幅度,即青少年未来的压力峰值或波谷,和其距离压力峰值或波谷的时间跨度。论文采用了基于统计和聚类的两种方法来确定模糊集的隶属度函数,并通过模糊决策树从模糊压力时间序列中发现压力变化规则,据此实现对压力变化幅度的预测。实验结果表明了基于模糊K线的预测模型的有效性。
[Abstract]:Teenagers have become one of the most stressful groups in society. Surveys show that young people in all countries are suffering from serious stress problems, and suicide has become an important cause of death among teenagers. Predicting the possible trend of psychological stress has become an important research issue. With the rapid development of social networks, more and more teenagers prefer to use this vivid rich media interaction to obtain information and self-expression. This makes it possible for us to obtain the psychological stress information of teenagers through Weibo. This paper predicts the psychological stress trends of teenagers through Weibo. These stress prediction results can serve as a warning to teenagers. Under pressure, teenagers can adjust their behavior in a timely manner and take reasonable measures to relieve the stress. To prevent the long-term accumulation of stress leading to serious psychological problems. Adolescent teachers and guardians can also better understand the stress situation of teenagers, provide timely personalized guidance, The main research work of this paper includes the following aspects: 1. Predict the psychological stress of teenagers at the next moment. Based on the data preprocessing work such as missing value filling, the multi-factor stress time series is obtained. In this paper, the multi-factor time series model is used to model the changing law of adolescent stress. Furthermore, considering the influence of events on adolescents' psychological stress, the change pattern of event stress is obtained through event trend mining, which modifies the prediction data of time series model and realizes more accurate prediction. By considering several stress factors and event correction, the prediction effect can be improved significantly. 2. To predict the change of psychological stress value of teenagers at the next moment. Based on the similarity between adolescent psychological pressure change and stock market price fluctuation, In this paper, the most commonly used K-line theory in stock trend analysis is introduced to express the pressure value information of teenagers, and through K-line based stress pattern matching, the change signal of pressure trend is captured from the historical K-line sequence. The prediction of the change trend of the pressure value at the next moment is realized, that is, the pressure increases. The experimental results show that the prediction error of the prediction model based on K line is 16.21, which is superior to other trend analysis methods .3.The range of psychological stress change of teenagers in the future is predicted. Combined with fuzzy data expressions to predict the range of future psychological stress among teenagers, that is, the peak or trough of future stress among teenagers. In this paper, two methods based on statistics and clustering are used to determine the membership function of fuzzy sets, and the rules of pressure change are found from fuzzy pressure time series by fuzzy decision tree. The experimental results show that the prediction model based on fuzzy K line is effective.
【学位授予单位】：清华大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：B844.2

【相似文献】