Big Data-Driven Precision Strategy Generation for College Admissions Using Gradient Boosted Decision Trees (GBDT)

As the scale of higher education enrollment expands and the complexity of enrollment decisions increases, traditional enrollment strategies have significant deficiencies in data analysis accuracy and resource allocation efficiency. This study proposes an enrollment data analysis and precise strategy generation system based on gradient boosting decision tree (GBDT) to solve the problem of insufficient data-driven decision-making in enrollment practice. By analyzing five years of enrollment data, the study used the GBDT model to achieve automatic screening and accurate prediction of high-dimensional variables. Experimental results show that the GBDT model is significantly better than the random forest and support vector machine models in performance indicators such as prediction accuracy, F1 score and AUC value, demonstrating excellent classification capabilities and high computational efficiency. Based on the model output, a number of optimization strategies were developed, including measures such as professional recommendation and regional enrollment focus adjustment. These strategies increased the overall enrollment success rate from 75% to 85%, significantly optimizing resource allocation efficiency and decision-making quality. The research innovation of this article is to combine the GBDT model to identify key variables and generate precise strategies, and to prove its practical application value in the field of higher education enrollment through experiments.