Scholarlink

This study introduces a machine learning approach to forecasting student dropout in South African schools, with a focus on identifying the main factors contributing to dropout rates. The research utilizes educational data and highlights variables such as the age at school entry, family participation, and absenteeism as important predictors. Several machine learning algorithms were evaluated, including Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Classifier (SVC), and XGBoost, with Random Forest being chosen for further refinement. After applying hyperparameter tuning through GridSearchCV, the Random Forest model showed strong performance across metrics like accuracy, precision, recall, and F1-score. The results illustrate the model’s effectiveness in accurately predicting dropout rates, serving as a valuable tool for the early detection of at-risk students. The findings support the integration of predictive analytics into school systems to track student progress and enable timely interventions. This research provides key insights for educators and policymakers working to reduce dropout rates using data-driven strategies.

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

A Machine Learning-Based Dropout Prediction Model for Rural and Township Schools in South Africa

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Njabulo Ndlela,
  • Sizakele Mathaba,
  • Alfredo Terzoli

摘要

This study introduces a machine learning approach to forecasting student dropout in South African schools, with a focus on identifying the main factors contributing to dropout rates. The research utilizes educational data and highlights variables such as the age at school entry, family participation, and absenteeism as important predictors. Several machine learning algorithms were evaluated, including Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Classifier (SVC), and XGBoost, with Random Forest being chosen for further refinement. After applying hyperparameter tuning through GridSearchCV, the Random Forest model showed strong performance across metrics like accuracy, precision, recall, and F1-score. The results illustrate the model’s effectiveness in accurately predicting dropout rates, serving as a valuable tool for the early detection of at-risk students. The findings support the integration of predictive analytics into school systems to track student progress and enable timely interventions. This research provides key insights for educators and policymakers working to reduce dropout rates using data-driven strategies.