Acad-GPS: Academic Grade Prediction System

Abstract

Students that fail courses have to retake those courses, often during regular semesters, sometimes during summer semesters. Every summer, NUST has to keep their departments open to o er courses to allow students to graduate from their programs without incurring excessive delays. The cost of engaging faculty to stay back for the summer is an additional nancial burden on universities' budgets. There is a cost to universities even when students retake courses during regular semesters. These losses and delays are a result of students failing courses because they underestimate the e ort they need to put in. In case of elective courses, students sometimes select courses that are not aligned with their inherent talents and abilities. In this research we propose to develop an academic grade prediction system (Acad-GPS), which predicts a student's future grades based on his/her academic history. This will allow students to prepare themselves for the academic rigour of upcoming courses. We have formulated this problem as a recommender system problem. The successful development of Acad-GPS will provide better guidance for university students, lead to fewer students failing courses, which will not only result in immediate cost savings to universities and the national exchequer, but also reduce average graduation times for students, avoid unnecessary delays of new entrants into the job market.