Scholarlink

<p>Education is vital to raising a generation that can lead a country to progress in all areas of life. The amalgamation and progression of machine learning, deep learning, and artificial neural networks allowed the development of models. Early performance prediction enables educators to effectively step in and assist students who are struggling or who might be at risk of falling behind and need additional support for academics. This study enables teachers an opportunity to transform their approach, identify areas to improve, and enhance student success significantly. Experiments were conducted using an OULAD (Open University Learning Analytics Dataset) from Kaggle, applying three machine learning models: Support Vector Machine (SVM), Categorical Boosting (CatBoost), Multilayer Perceptron (MLP), and Proposed Ensemble Model (PEM) using a soft voting approach. Among these models, our PEM emerged as the best performer and generalized effectively on the dataset, with accuracy rates of 96.83%. Therefore, it could potentially be stated that deeper and more optimized frameworks are more suitable for the task of classification of students’ grades. Based on the current study, deep learning can automate grades for early outcome prediction that helps educators, policymakers, and students who struggle in the moment. In addition, it increases academic success, identifies areas for improvement, and transforms teaching and learning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Blended Ensemble Intelligence for Academic Performance Prediction in Educational Datasets

  • 期刊论文

  • 发表日期：

    2026年3月25日

查看全文
  • Minakshi Sharma,
  • Vibhakar Mansotra

摘要

Education is vital to raising a generation that can lead a country to progress in all areas of life. The amalgamation and progression of machine learning, deep learning, and artificial neural networks allowed the development of models. Early performance prediction enables educators to effectively step in and assist students who are struggling or who might be at risk of falling behind and need additional support for academics. This study enables teachers an opportunity to transform their approach, identify areas to improve, and enhance student success significantly. Experiments were conducted using an OULAD (Open University Learning Analytics Dataset) from Kaggle, applying three machine learning models: Support Vector Machine (SVM), Categorical Boosting (CatBoost), Multilayer Perceptron (MLP), and Proposed Ensemble Model (PEM) using a soft voting approach. Among these models, our PEM emerged as the best performer and generalized effectively on the dataset, with accuracy rates of 96.83%. Therefore, it could potentially be stated that deeper and more optimized frameworks are more suitable for the task of classification of students’ grades. Based on the current study, deep learning can automate grades for early outcome prediction that helps educators, policymakers, and students who struggle in the moment. In addition, it increases academic success, identifies areas for improvement, and transforms teaching and learning.