Comparative Study of Machine Learning Algorithms for Student Retention, early warning and intervention systems for Institutions of Higher Learning

The main purpose of this research work is to conduct a comparative study of application of machine learning algorithms in the development of a student retention model for undergraduate students at an institution of higher learning. The Decision Tree, Random Forest, K-Nearest Neighbor, Gradient Boosting Trees, Logistic Regression, Stochastic Gradient Descent, Linear Support Vector Classification and Gaussian Naïve Bayes are used in the development of the Undergraduate student retention model. The implementation results show that The Decision Tree, Random Forest, K-Nearest Neighbor, Gradient Boosting Trees and Logistic Regression achieved an accuracy of ${9 8 . 5 9} \%, {9 4 . 1 1} \%, {8 9 . 5 5} \%$ and ${8 9 . 4 2} \%$ respectively. To improve this work in future it is recommended that the current machine learning algorithms could be amalgamated with Artificial Neural Networks or deep learning algorithms for improved performance. Good features identified in traditional models used for development of student retention models can also be integrated with machine learning models for improved performance.