Data Selection, Training, and Validation for Deployment of the Artificial Neural Networks to Predict Science Education Students’ Early Completion of University

Abstract

Predicting student graduation helps schools identify students in danger of dropping out and intervene early to enhance academic performance. It can also assist educators and policymakers in creating graduation and dropout prevention initiatives. Nowadays, predicting students' performance is one of the most specific topics for learning environments, such as universities and schools, since it leads to the development of effective mechanisms that can enhance academic outcomes and avoid destruction. This study aims to (1) determine the datasets needed for the adaption/simulation of fresh undergraduate science education students across their area of specialization from the 2014/2015 to the 2017/2018 academic session and (2) determine the artificial neural network process of training, testing, and validation of datasets for science education students’ graduation. The artificial neural network that was adapted was able to predict the early completion of the student with an accuracy of 99.9% based on tests conducted on 417 datasets. The result of the study indicated that the early cumulative grade point average (CGPA) for three consecutive semesters was necessary for the adaptation of artificial neural network simulation. The ANN accepted the training, testing, and validation of the dataset by randomly separating the data into 70%, 15%, and 15%, respectively. The result of the study indicated that grade point average (GPA) for three consecutive semesters (GPA1, GPA2, and GPA3) is much needed for the ANN simulation. There should be strategic interventions from the university authority to monitor student performance at an early stage. The authority should introduce the ANN predicting model into the department, which could help lecturers to systematically monitor student performance.