MOOC学习行为数据的高阶离散MPT模型分析

Analysis on Higher-Order Discrete MPT Model of MOOC Learning Behavior Data

文章在泊松边际分布条件下,将传统的一阶混合Thinning和Pegram算子的MPT离散值时间序列模型拓展到高阶MPT(p)模型。研究了模型的自相关结构,分析了模型阶数与参数估计的方法,推断了参数性质和回归性质,给出了模型的预测方法。通过对MOOC课程的学习行为离散数据序列进行建模分析,描述了在线学习者的行为特征,发现了在线学习行为高阶滞后的相依影响,实现了离散序列的短期预测。

Under the condition of Poisson marginal distribution, this paper develops MPT discrete values of the time series model based on the traditional first-order mixed Thinning and Pegram operators into a new higher-order MPT(p) model. The paper firstly studies the autocorrelation structure and analyzes the methods of model order and parameter estimation, and then deduces the properties of parameters and regression, and presents the prediction method of the model. Finally, by modeling the discrete data sequence of the MOOC course’s learning behavior, the paper describes online learners’ behavior characteristics and discovers the dependent influence of higher order delay of online learning behavior, realizing the short-term prediction of discrete sequence.