Scholarlink

<p>Predicting student performance is vital to educational data mining. The design and development of rapid, efficient, and accurate methods for predicting student achievement are the keystones of this field. Predicting student performance is an essential indicator of students’ understanding of course material and instructors’ teaching effectiveness. Accurate prediction of student performance not only helps provide timely feedback to students and educators but also assists university administrators in assessing course quality. This article presents a hybridization approach to predicting student achievement, combining the Weighted Extreme Learning Machine with the Coati Optimization Algorithm (COA). The robustness of the prediction model is established and formulated using an Extreme Learning Machine (ELM) and a Weighted Extreme Learning Machine (WELM), with a thorough investigation of the interplay between courses and student attributes. The biases and weights in the ELM and WELM architectures are randomly initialized, which may lead to inconsistent results, resulting in training errors and a decline in prediction accuracy. Hence, to solve this issue, the COA is combined with WELM to get optimal input weights and hidden biases, thereby enabling WELM to achieve greater accuracy. The predictive capabilities of the implemented models are evaluated using two standard student datasets, and their effectiveness is compared with that of the Spike Neural Network (SNN) algorithm. The experimental results indicated that WELM-COA outperformed SNN.</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

Data-Driven Student Performance Prediction: A Hybrid Approach Using Weighted Extreme Learning Machine and Coati Optimization Algorithms

  • 期刊论文

  • 发表日期：

    2026年1月7日

查看全文
  • T. Bharathi,
  • R. Manavalan

摘要

Predicting student performance is vital to educational data mining. The design and development of rapid, efficient, and accurate methods for predicting student achievement are the keystones of this field. Predicting student performance is an essential indicator of students’ understanding of course material and instructors’ teaching effectiveness. Accurate prediction of student performance not only helps provide timely feedback to students and educators but also assists university administrators in assessing course quality. This article presents a hybridization approach to predicting student achievement, combining the Weighted Extreme Learning Machine with the Coati Optimization Algorithm (COA). The robustness of the prediction model is established and formulated using an Extreme Learning Machine (ELM) and a Weighted Extreme Learning Machine (WELM), with a thorough investigation of the interplay between courses and student attributes. The biases and weights in the ELM and WELM architectures are randomly initialized, which may lead to inconsistent results, resulting in training errors and a decline in prediction accuracy. Hence, to solve this issue, the COA is combined with WELM to get optimal input weights and hidden biases, thereby enabling WELM to achieve greater accuracy. The predictive capabilities of the implemented models are evaluated using two standard student datasets, and their effectiveness is compared with that of the Spike Neural Network (SNN) algorithm. The experimental results indicated that WELM-COA outperformed SNN.