Systematic Analysis of Student Academic Performance Prediction: Prediction Model, Optimization, Data Augmentation, and Educational Policy Implications

Predicting student academic performance is a critical challenge in higher education that aims to improve the quality of learning and optimize support for students. The main issues faced are how to choose the right prediction model, utilize optimization methods, and improve prediction accuracy. This paper presents a systematic review of the literature related to various prediction models, ranging from traditional statistical models to machine learning and ensemble/hybrid models, and discusses optimization techniques such as evolutionary algorithms. In addition, determinants of academic performance such as motivation, student engagement, and institutional support are identified as important variables in predicting learning outcomes. The datasets used, model evaluation methods, and technologies and tools that support the analysis process are also discussed in detail. The results of this review show that a hybrid approach that combines machine learning and optimization techniques is able to provide more accurate predictions compared to traditional models. This paper also highlights existing limitations, including challenges in model generalization, and provides policy recommendations to improve the effectiveness of academic performance prediction and its implementation in educational practice.