基于深度学习和行为数据的自控力评价研究

Research on Self-Control Evaluation Based on Deep Learning and Behavioral Data

自我控制能力(自控力)是个体适应社会、达成人生目标的一项重要能力。但目前心理学常用的任务实验和量表等方法需要消耗大量的时间成本和人力资源,同时研究对象的主观意识会影响评价结果的置信度。文章提出了一种基于深度学习LSTM(Long Short-Term Memory)模型的自控力量化评价方法,该方法以大学生在校园内的一卡通行为数据为数据源,利用LSTM模型提取行为数据时序特征,在测试数据集上自控力评价结果与心理学量表法测量结果的相关系数为0.66,且能够消除研究对象的主观影响,适合学校进行大规模的学生自控力测评,具有重要的应用前景。

Self-control ability(self-control)is an important ability for individuals to adapt to society and achieve life goals.However,the task experiments and scales commonly used in psychology need to consume a lot of time and human resources.At the same time,the subjective consciousness of the research object will affect the confidence of the evaluation results.This paper proposes a self-control evaluation method based on LSTM model.The method uses the behavior data of college students on campus as the data source,and uses the LSTM model to extract the time series characteristics of the behavior data.On the test set,the correlation coefficient between the evaluation result and the psychological scale method measurement result is 0.66,and the method can eliminate the subjective influence of the research object.It is suitable for large-scale student self-control evaluation in schools and has important application prospects.