Predicting students’ results in higher education using a neural network

A significant problem in higher education is the poor results of students after admission. Many students leave universities from a variety of reasons: poor background knowledge in the field of study, very low grades and the incapacity of passing an examination, lack of financial resources. Predicting students’ results is an important problem for the management of the universities who want to avoid the phenomenon of early school leaving. We used a neural network to predict the students’ results measured by the grade point average in the first year of study. For this purpose we used a sample of 1000 students from “Nicolae Titulescu” University of Bucharest from the last three graduates’ generations, 800 being used for training the network and 200 for testing the network. The neural network was a multilayer perceptron (MLP) with one input layer, two hidden layers and one output layer and it was trained using a version of the resilient backpropagation algorithm. The input data were the students profile at the time of enrolling at the university including information about the student age, the GPA at high school graduation, the gap between high school graduation and higher education enrolling. After training the network we obtained MSE of about 1.7%. The ability to predict students’ results is of great help for the university management in order to take early action to avoid the phenomenon of leaving education.