Chen Yanyu, Yao Xiaolin
School of Education, Durham University, Leazes Road, Durham, DH1 1TA, UK.
School of Information and Business Management, Dalian Neusoft University of Information, Dalian, 116021, Liaoning, China.
Sci Rep. 2025 Mar 28;15(1):10809. doi: 10.1038/s41598-025-86261-y.
Academic achievement is vital for campus life and education since it indicates the caliber of the teachers, administration, and students' learning abilities. Issues such as poor study conditions and family disruptions can impede a student's capacity to achieve. Teachers are looking for practical solutions to these concerns because solving problems one at a time might be tough. This study uses a combination of black hole optimization (BHO) and Gaussian process regression (GPR) algorithms to predict students' academic success in higher education. The method is divided into three stages: data pre-processing, identification of effective indicators using BHO algorithms, and forecasting of academic performance. The presented approach makes use of the GPR algorithm to choose the relevant features and the weighted combination of GPR models to forecast that the GPR model would be used for the weighting operation that is, to determine the ideal weights. The experimental findings demonstrate that our method has a lower error rate of 0.95 and 0.81 in terms of RMSE and MAE than the competing methods. The proposed method can assist teachers in analyzing student behavioral patterns, understanding academic performance impact mechanisms, and developing effective learning supervision plans.
学业成绩对校园生活和教育至关重要，因为它能体现教师、管理人员的水平以及学生的学习能力。诸如学习条件差和家庭变故等问题会阻碍学生取得好成绩的能力。教师们正在寻找解决这些问题的切实可行的办法，因为逐个解决问题可能很困难。本研究结合黑洞优化（BHO）算法和高斯过程回归（GPR）算法来预测高等教育中学生的学业成就。该方法分为三个阶段：数据预处理、使用BHO算法识别有效指标以及预测学业成绩。所提出的方法利用GPR算法选择相关特征，并通过GPR模型的加权组合来预测将用于加权操作的GPR模型，即确定理想权重。实验结果表明，我们的方法在均方根误差（RMSE）和平均绝对误差（MAE）方面的错误率分别为0.95和0.81，低于竞争方法。所提出的方法可以帮助教师分析学生的行为模式，理解学业成绩的影响机制，并制定有效的学习监督计划。
