A Performance Prediction Model for Newly Enrolled University Students Based on The Fusion of K-Means and LightGBM Algorithms

Accurately predicting the academic performance (GPA) of newly admitted university students is crucial for educational management and resource allocation. To improve prediction accuracy, this study proposes a fusion algorithm that combines K-means clustering and LightGBM regression. The model utilizes multi-dimensional student features, including province, gender, region, entrance exam scores, and high school subject selection, to predict their GPA. First, the K-Means algorithm is applied to cluster the student population, grouping students into several clusters to capture heterogeneity among them. Second, the resulting clusters are then integrated into the LightGBM model as new features for regression-based performance prediction, further enhancing accuracy. Experimental results show that the fusion model of K-means and LightGBM performs well across different values of K, with optimal results achieved at <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">K=5</tex>. This approach improves the model's generalization and applicability, providing educational administrators with a basis for personalized interventions for new students.