基于分层线性模型的学生数据挖掘

Based on the investigation of the independent variables commonly used in education data mining and the official analysis report of National Entrance Examination of College(NEEC) and Entrance Examination of High School(EEHS), this thesis establishes indicators that measure student skills change from junior high school to high school and call it the Improvement of Student Skills(ISS). At present most of the independent variables used in the analysis are students\' personal information and test scores, different aspect of student ability examined by test paper is not fully studied. Based on the hierarchical linear model, following questions are studied: 1) Influence that ISS and school features have on prediction of student NEEC performance; 2) Reflecting different influence on ISS among schools. As a result, ISS can describe student better than simply student raw score, improve theprediction result. Apart from that, the model can also display the differen influence that school had on studendt ISS. By comparing two school, This thesis find that the two school each have its own advantages on positive influence on student ISS, however, the school that rank higher has smaller negative influence on student ISS.