Research and application of university students’ comprehensive management decision system based on data visualization

Abstract

The student management in colleges and universities is facing unprecedented complexity and challenges, which is caused by the continuous development of higher education and the rapid advancement of digitalization. In this study, a comprehensive management decision-making system integrating data visualization is constructed to improve the scientificity and accuracy of student management in higher education institutions. The system uses the upgraded random forest (RF) model to implement academic early warning, the updated K-means clustering method to identify the rules of students’ behavior, and a multi-dimensional visual dashboard to integrate and display the analysis data. According to the experimental results, the improved RF model is superior to the conventional RF model and logistic regression model in predicting high-risk students, with an accuracy rate of 91.8% and a F1 score of 91.0%. The silhouette coefficient of the improved K-means clustering is between 0.64 and 0.71. Students are successfully divided into three groups: top-performing students, well-rounded students and procrastinating students. The comprehensive visual dashboard intuitively presents the trends of academic risks and the distribution of behavior-based student groups, providing managers with scientific and actionable decision support. Based on the research findings, this approach can be used to improve student management effectiveness and develop intervention techniques.