Identifying Academically At-Risk Students Using Predictive Analysis Model

The increasing dropout rates in higher education institutions underscore the critical need for proactive strategies to identify academically at-risk students.This study presents the development and evaluation of a predictive analysis model leveraging machine learning-specifically the Random Forest algorithm-to accurately identify students at risk of academic failure.The model integrates both academic indicators (e.g., GPA, attendance, exam scores) and non-academic factors (e.g., socio-economic status, behavioral patterns, family dynamics) to provide a holistic assessment of student performance.A dataset of 100,256 student records from the Australian Student Performance Dataset was preprocessed, with key features selected to enhance model accuracy.The model achieved a predictive accuracy of 69% and was deployed through a webbased application developed using the Flask framework.Functionality includes real-time prediction, risk classification, and user-friendly visualization.Stakeholder evaluation involving 40 respondents showed 88% user satisfaction, confirming the system's reliability, usability, and practical value.The findings demonstrate the model's effectiveness in enabling early interventions, thereby contributing to reduced attrition rates and more inclusive, data-informed educational practices.