Scholarlink

This paper addresses the challenge of class imbalance in student dropout prediction using boosting algorithms combined with Synthetic Minority Oversampling Technique (SMOTE). The study evaluates five boosting algorithms (XGBoost, LightGBM, CatBoost, AdaBoost, and Gradient Boosting) on both original and SMOTE-balanced datasets. Using a dataset of 4,424 student records with 36 features, experimental results demonstrate that SMOTE significantly improves model performance. LightGBM achieved the best results on SMOTE-balanced data with 81.39% accuracy and 0.9352 ROC-AUC score, showing an improvement of 4.21 percentage points over the original dataset. The findings suggest that combining SMOTE with boosting algorithms effectively enhances the identification of at-risk students in higher education.

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

Boosting Academic Success: Predictive Modeling on Imbalanced Student Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Hassan Silkhi,
  • Brahim Bakkas,
  • Khalid Housni

摘要

This paper addresses the challenge of class imbalance in student dropout prediction using boosting algorithms combined with Synthetic Minority Oversampling Technique (SMOTE). The study evaluates five boosting algorithms (XGBoost, LightGBM, CatBoost, AdaBoost, and Gradient Boosting) on both original and SMOTE-balanced datasets. Using a dataset of 4,424 student records with 36 features, experimental results demonstrate that SMOTE significantly improves model performance. LightGBM achieved the best results on SMOTE-balanced data with 81.39% accuracy and 0.9352 ROC-AUC score, showing an improvement of 4.21 percentage points over the original dataset. The findings suggest that combining SMOTE with boosting algorithms effectively enhances the identification of at-risk students in higher education.