Scholarlink

This study investigates methods for predicting student performance and identifying potential dropouts using advanced machine learning algorithms. The research aims to improve early intervention strategies by analyzing a range of academic, demographic, and behavioral data. A comprehensive dataset, including grades, attendance records, and socio-economic indicators, was processed using multiple predictive models such as logistic regression, K–nearest neighbors, support vector machines, decision tree, random forest, XGBoost, and neural networks. The results demonstrate that key features like academic performance and engagement levels are highly indicative of future outcomes. The best-performing model achieved an accuracy of over 85%, significantly improving early dropout detection rates. Our findings suggest that integrating predictive analytics into educational systems can provide actionable insights for administrators and educators, enabling targeted support for at-risk students. This research reached its conclusion by analyzing the outcomes for policy formulation and upcoming research in education technology and customized learning approaches.

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

Student Dropout Forecast and Evaluation Using Multiple Machine Learning Models

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Yuvraj Singh,
  • Rishabh Dev Singh,
  • Nilamadhab Mishra,
  • Tiansheng Yang,
  • Danyu Mo,
  • Bharati Rathore

摘要

This study investigates methods for predicting student performance and identifying potential dropouts using advanced machine learning algorithms. The research aims to improve early intervention strategies by analyzing a range of academic, demographic, and behavioral data. A comprehensive dataset, including grades, attendance records, and socio-economic indicators, was processed using multiple predictive models such as logistic regression, K–nearest neighbors, support vector machines, decision tree, random forest, XGBoost, and neural networks. The results demonstrate that key features like academic performance and engagement levels are highly indicative of future outcomes. The best-performing model achieved an accuracy of over 85%, significantly improving early dropout detection rates. Our findings suggest that integrating predictive analytics into educational systems can provide actionable insights for administrators and educators, enabling targeted support for at-risk students. This research reached its conclusion by analyzing the outcomes for policy formulation and upcoming research in education technology and customized learning approaches.