Scholarlink

<p>Artificial Intelligence (AI) has emerged as a rapidly influencing element in higher education, reshaping traditional approaches to instruction, student learning, and institutional management. This study introduces an integrated AI framework that integrates hybrid machine-learning methods with deep learning (DL) models such as XGBoost, Random Forest, TabNet, and BERT to forecast student outcomes. The dataset incorporates variables related to demographics, academics, and learner behavior, which undergo rigorous preprocessing steps including imputation, normalization, and class balancing. Comparative results indicate that BERT and TabNet provide higher prediction accuracy and better generalization across diverse datasets compared to conventional ensemble models. The proposed framework effectively integrates structured and unstructured data along with multiple evaluation metrics to enhance model interpretability and stability. The findings highlight the usefulness of early-alert tools and data-guided interventions that improve institutional efficiency and student outcomes. Leveraging responsible AI enables universities and colleges to achieve new levels of innovation, personalization, and operational effectiveness, signaling a shift toward a more flexible and student-centered educational environment.</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

AI-Driven predictive analytics for student success and institutional decision-making in higher education

  • 期刊论文

  • 发表日期：

    2026年1月14日

查看全文
  • Fadheela Hussain,
  • Mustafa Hammad,
  • Haitham Isa Al Qahtani

摘要

Artificial Intelligence (AI) has emerged as a rapidly influencing element in higher education, reshaping traditional approaches to instruction, student learning, and institutional management. This study introduces an integrated AI framework that integrates hybrid machine-learning methods with deep learning (DL) models such as XGBoost, Random Forest, TabNet, and BERT to forecast student outcomes. The dataset incorporates variables related to demographics, academics, and learner behavior, which undergo rigorous preprocessing steps including imputation, normalization, and class balancing. Comparative results indicate that BERT and TabNet provide higher prediction accuracy and better generalization across diverse datasets compared to conventional ensemble models. The proposed framework effectively integrates structured and unstructured data along with multiple evaluation metrics to enhance model interpretability and stability. The findings highlight the usefulness of early-alert tools and data-guided interventions that improve institutional efficiency and student outcomes. Leveraging responsible AI enables universities and colleges to achieve new levels of innovation, personalization, and operational effectiveness, signaling a shift toward a more flexible and student-centered educational environment.