基于时序聚类和集成学习的用户学习模式挖掘与识别

In this paper, k-means for longitudinal data (KmL) and ensemble learning are adopted to analyze the users\' learning behavior of a certain English vocabulary APP of China.First of all, the KmL algorithm is adopted to cluster thetrajectories of words of users, and two typical learning patterns are obtained.Then, based on the feature selection algorithm and the prediction results of the base model, the representational attributes of the learning state which can be used to predict the learning pattern are selected.Finally, stacking combine stratery was used for the final integration of a variety of single classifiers and homogeneous ensemble classifiers, and the user learning pattern prediction model was obtained.The experimental results show that according to the learner\'s learning state attribute, the category of the learning patterns can be predicted, and then the trend of the user\'s memorization volume and the completion degree of the word book can be predicted.The stacking heterogeneous ensemble model with SVM as the meta-classifier shows high accuracy in this research question.