【作 者】：
王希哲， 黄昌勤， 朱 佳， 徐小琪
【关 键 词 】：
网络学习空间； 学习云空间； 教育大数据； 关联性判别； 学情预测； 并行随机森林
【栏 目】：
网络学习空间
【中文摘要】：
随着学习云空间应用的不断深入，海量过程性数据与富媒体资源使其呈现为典型大数据特征环境。文章对学习云空间的大数据特征、分析方法及其带来的机遇进行了深入的分析，并依据学习云空间传统数据及其过程性数据提出一种学情预测模型及其实现方法。在模型构建过程中，通过计算各学情因素的Gini增益，实现学习效果影响程度的判别，并提出改进的并行随机森林算法，以世界大学城系统平台为支撑进行学习预测与效果检验。结果表明，该方法较为有效地实现了大数据环境下学习云空间的学情预测，并为学习云空间中基于大数据技术的智能服务提供了一种可行的参考方案。
【英文摘要】：
With the deepening application of cloud learning space, the cloud learning space is becoming a typical environment of big data because of massive process data and rich media resources. This paper firstly analyzes the characteristics, analysis methods and opportunities of big data in cloud learning space, and then proposes a learning condition prediction model and its implementation method based on the traditional data and process data in cloud learning space. During the model building, the Gini Gain of every learning condition factor is calculated to distinguish their impacts on learning effect. After that, an improved parallel algorithm of Random Forests is purposed to predict learning situation and verify its effects in WordUC platform. The results show that this method can effectively predict the learning situation of cloud learning space in big data environment and provide a feasible reference scheme for big data-based intelligent service in cloud learning space.