A Simple Framework for Predicting Student Dropout Analysis Using Data Mining Algorithms

Dropout is an essential concern in the educational sector with implications for students as well as the institutions. This paper proposed a predictive model that uses machine learning techniques to identify at-risk students for dropout. Four years of data were used for 270 students on an online LMS by using different models such as Logistic Regression, Decision Tree, Naive Bayes, SVM, Random Forest, and ANN. The key features like LMS access, grades, and performance of assignments were evaluated after the data preprocessing involved cleaning and one-hot encoding. AUC measures were used for evaluating the performance of the model. Naive Bayes turned out to be the most effective model and followed by SVM. The results demonstrate the utility of machine learning for early dropout prediction, equipping educators with an important resource for intervention to help atrisk students.