基于一卡通消费数据的学生成绩预测和朋友关系网络检测研究

发布时间：2018-10-29 13:48

【摘要】：随着信息化与数字化校园建设的不断深入发展,校园“一卡通”系统的应用已经渗入到高校大学生的学习、生活等方方面面。在这种背景下,大量的学生的各种活动“痕迹”数据都以文本的方式被完整地记录下来,为我们分析各种学生活动提供了可能。如果我们能利用获得的海量数据,挖掘出学生的行为特点,将无疑为学校的日常管理和决策制定提供重要参考。因此怎样利用这些数据为完善校园相关方面的建设提供有利帮助,是校园卡数据研究的重点方向。目前关于一卡通消费数据的研究集中在通过统计或者数据挖掘的方法来研究学生的消费行为。而大学生在校期间的主要时间是在校内学习和生活,这决定了大学生最重要的行为是“学习行为”和“交往行为”。传统上通过观察、调查问卷等方法来研究学生的行为,但是由于数据的客观性差、数据量小等原因,使得这一问题很难得到满意的解决。能否通过一卡通消费数据来研究学生的“学习行为”和“交往行为”是我们关心的问题。而“学习成绩”和“朋友关系”分别作为“学习行为”和“交往行为”的直接结果,是对它们的重要反映。因此,如何利用一卡通消费数据研究学生的“学习成绩”和“朋友关系”是研究学生的“学习行为”和“交往行为”的重要研究方向。本文基于一卡通消费数据,借鉴了复杂网络的研究方法研究了学生的“学习成绩”和“朋友关系”。本文主要工作如下：(1)细致分析了学生早餐次数与专业成绩之间的秩相关性以及消费时间间隔序列阵发性与专业成绩的秩相关性,并使用机器学习的KNN (K Nearest Neighbor)分类算法对学生的成绩等级进行了预测,取得了很高的准确率。(2)基于一卡通消费数据构建了消费数据的二分网络,并采用多重检验的方法对学生共现是否源于随机相遇进行了统计验证,得到了学生朋友关系网络,然后对所得到的学生朋友关系网络的网络基本特征进行了分析。
[Abstract]:With the continuous development of information and digital campus construction, the application of campus "one-card" system has penetrated into the study, life and other aspects of college students. In this context, a large number of student activities "trace" data are recorded in the form of text, which provides us with the possibility to analyze various student activities. If we can make use of the massive data we get, we can find out the characteristics of students' behavior, which will undoubtedly provide an important reference for the daily management and decision-making of the school. Therefore, how to make use of these data for the improvement of campus related aspects of the construction is the key direction of campus card data research. At present, the research on the consumption data of one card is focused on the research of students' consumption behavior by means of statistics or data mining. The main time of college students is to study and live in school, which determines that the most important behaviors of college students are "learning behavior" and "communication behavior". Traditionally, students' behavior is studied by means of observation, questionnaire and so on. However, because of the poor objectivity of data and the small amount of data, it is difficult to solve this problem satisfactorily. Whether we can study the students'"learning behavior" and "communication behavior" through the consumption data of one-card is our concern. As the direct result of "learning behavior" and "communication behavior", "academic achievement" and "friend relationship" are important reflection of them. Therefore, it is an important research direction to study students'"learning behavior" and "communication behavior" by using the data of one-card consumption to study students'"academic achievement" and "friends' relationship". Based on the one-card consumption data, this paper studies students'"learning achievement" and "friend relationship" by using the research method of complex network. The main work of this paper is as follows: (1) the rank correlation between breakfast frequency and professional achievement and the rank correlation between paroxysmal consumption time series and professional achievement are analyzed in detail. The KNN (K Nearest Neighbor) classification algorithm of machine learning is used to predict the grade of students, and a high accuracy rate is obtained. (2) based on the consumption data of one card, a binary network of consumption data is constructed. The method of multiple test is used to verify whether the co-occurrence of students originated from random encounter, and the network of student friend relationship is obtained. Then, the basic characteristics of the network of students' friend relationship are analyzed.
【学位授予单位】：华中师范大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.13;TN409

【相似文献】