Student Consultancy Service: Prediction of Course Grades in Course Selection Phases Using Artificial Intelligence Techniques

Abstract

Universities offer technical elective courses to allow students to improve themselves in various parts of their majors. Each semester, the students make a decision regarding these technical electives, and the most common expectations students have in this context include, getting education at a better school, getting a better job, and getting higher grades with a view to securing admission into more advanced degree programs. Electing a course on the basis of the interests and skills of the student will naturally translate into achievement. Advisors, in this context, play a major role. Yet, the substantial workload advisors have already assumed prevent them dedicating enough time for exploring the interests and skills of the students, and hence hinder the development of the required relationship between students and their advisors. This study attempts to estimate the achievement level a student intends to elect, on the basis of graduate data received from the database of students of Sakarya University, Faculty of Computer and Information Sciences, and led to the development of a decision-support system. The application used ANFIS and artificial neural network methods among the artificial intelligence techniques, alongside the linear regression model as the mathematical model, whereupon the performance of the methods were compared over the application. In conclusion, it was observed that artificial intelligence techniques provided more relevant results compared to mathematical models, and that, among the artificial intelligence techniques feed forward backpropagation neural network model offered a lower standard deviation compared to ANFIS model.