Predicted Specified STEAM Student Majors Depending On Many Factors Using Generative Adversarial Networks

Abstract

The traditional educational systems in certain nations, such as those in the Arab world, use the previous year's scores to forecast academic achievement. Meanwhile, the science, technology, engineering, arts and mathematics (STEAM) educational system also considers the student's talents and interests, in addition to their test results, for each of the years of study used to forecast academic achievement. However, despite the numerous variables that could potentially impact a student's future profession, the STEAM educational system accurately predicts the academic achievement of students in five to seven broad majors. A smart model, data, generative, factors (DGF), has been proposed to assign a specific major for each student based on their academic background, interests, skills, and main influencing factors. The DGF model uses the supervised deep neural network and is trained by the generative adversarial network (GAN) algorithm. The DGF model has the capability to predict a specific major among 17 different majors for every student, with a high learning performance (1.4133), a plausible error rate (0.0046), a rational number of learning epochs (206), and an optimal accuracy of 99.99%.