Minimizing Student Failure with Machine Learning in Computer Science and Engineering Classes

In 2022, 41.9 million Americans dropped out of college, highlighting a critical problem in the higher education sector. The causes of such high dropout rates are not limited and diverse, including academic struggles, financial constraints, family issues, among others. The early identification of students who may drop out and intervention are necessary to address these issues and ensure students' success in their educational journey. This research presents a machine learning-based approach to identify students at risk of failing the course. Our findings indicate that XGBoost with the GridSearchCV fine-tuning method achieve an accuracy of $\text{90.8 \%}$ to identify at-risk students and 73.4 % to predict the students' grades. These results reinforce the potential of data-driven approaches to mitigate dropout rates and enhance student success.