Probation Status Prediction and Optimization for Undergraduate Engineering Students

Abstract

The student performance monitoring process is essential because it can rapidly help students who have problems studying before they fail during the semester, causing them to be retired and impacting institutes. Thus, this research was conducted to analyze student performance toward the admission system to predict student probation status respected to other factors before ending the semester to help students. The research also conducted prescriptive analytics to optimize factors that may impact students' probation status using the evolutionary optimization algorithm. This analytics aims to generate an action plan for monitoring student characteristics that may fail and improve the educational process and support admission strategic planning before recruiting students. The five machine learning algorithms are used in the research consists of Logistic Regression, Deep Learning, Decision Tree, Random Forrest, and Gradient Boosted Tree. The model that gives the highest accuracy is GBT, which gives 96.2% and is chosen to use in prescriptive analytics, giving the action plan for the institutes.