Research on Diagnostic Test and Treatment for Higher Education System

This paper proposes a method to build machine learning models based on data analysis, which aims at providing strategies to evaluate the health and sustainability of the higher education system and the improvement measures correspondingly. Initially, the Key Performance Indicators (KPIs) method is used to judge the performance and operation status of the national higher education system. Factor Analysis is used to test collinearity and reduce dimension of these indicators, and the main indexes in BACKGROUND, INPUT, PROCESS, OUTPUT and utilization rate are screened out. Data Envelopment Analysis (DEA) is used to establish a higher education evaluation system suitable for countries and regions, and the scores of Germany's higher education system are calculated in five dimensions respectively. Secondly, by using Markov Chain Model for prediction, the periodic changing rule of German educational reform is gained. Combined with the knowledge of management policy, the time arrangement of the specific implementation of policies is obtained. Due to the advantages of DEA in evaluating the relative effectiveness of multi-input and multi-output systems, an improved GP-DEA model is constructed, based on C
²
R of DEA method and a target programming model is built, to quantify the effectiveness Finally, a sensitivity analysis is done, illustrating that this model has high efficiency and only has a small impact on its performance if the model hypothesis has a small deviation.