TY - GEN
T1 - Maximizing Grade Point Average Through Optimal Course Recommendations
T2 - 3rd International Conference on Advances in Data-driven Computing and Intelligent Systems, ADCIS 2024
AU - Srivats, R.
AU - Jackson, Harsha
AU - Mohapatra, Rajesh Kumar
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.
PY - 2026/1/1
Y1 - 2026/1/1
N2 - In the modern educational landscape, students choose a wide variety of courses from their university’s curriculum, each of which could have an impact on their academic career. As these decision-making processes are quite complex, they seek recommendations that address both their personal strengths and weaknesses. This paper introduces a novel approach to a course recommender aimed at optimizing a student’s Grade Point Average (GPA) using the Thompson Sampling Algorithm. The proposed model offers a complete solution by recommending a well-chosen set of courses that not only leverage students’ strengths but also consider their weaknesses, reducing the possibility of adverse GPA outcomes. The study's motive is to provide a personalized, data-driven approach to help overwhelmed students choose courses from their extensive curriculum, facilitating their academic success. Thompson Sampling—a technique within the Multi-Armed Bandits framework, also called as an exploration–exploitation technique, allows the model to recommend courses based on a student's prior grades. By employing a probability distribution for each course, the algorithm balances the exploration of unregistered courses and the exploitation of well-performing ones, resulting in the best possible recommendations. The model's ability to cater to each student's distinct abilities, preferences, and aspirations is what makes it innovative. The approach provides a balanced academic journey by guiding students through courses that align with their strengths along with courses that target their deficiencies. The paper provides an alternate methodology for students to stop depending on external biased opinions on recommended courses and instead offer a customized, data-backed approach. The proposed model achieves an accuracy of 84.8%, a precision of 87.12%, a recall of 89%, and an F1-score of 88%, surpassing existing methods. This encourages a holistic academic experience while empowering students to make decisions that reflect their potential and are well-informed, thereby improving their grade point average.
AB - In the modern educational landscape, students choose a wide variety of courses from their university’s curriculum, each of which could have an impact on their academic career. As these decision-making processes are quite complex, they seek recommendations that address both their personal strengths and weaknesses. This paper introduces a novel approach to a course recommender aimed at optimizing a student’s Grade Point Average (GPA) using the Thompson Sampling Algorithm. The proposed model offers a complete solution by recommending a well-chosen set of courses that not only leverage students’ strengths but also consider their weaknesses, reducing the possibility of adverse GPA outcomes. The study's motive is to provide a personalized, data-driven approach to help overwhelmed students choose courses from their extensive curriculum, facilitating their academic success. Thompson Sampling—a technique within the Multi-Armed Bandits framework, also called as an exploration–exploitation technique, allows the model to recommend courses based on a student's prior grades. By employing a probability distribution for each course, the algorithm balances the exploration of unregistered courses and the exploitation of well-performing ones, resulting in the best possible recommendations. The model's ability to cater to each student's distinct abilities, preferences, and aspirations is what makes it innovative. The approach provides a balanced academic journey by guiding students through courses that align with their strengths along with courses that target their deficiencies. The paper provides an alternate methodology for students to stop depending on external biased opinions on recommended courses and instead offer a customized, data-backed approach. The proposed model achieves an accuracy of 84.8%, a precision of 87.12%, a recall of 89%, and an F1-score of 88%, surpassing existing methods. This encourages a holistic academic experience while empowering students to make decisions that reflect their potential and are well-informed, thereby improving their grade point average.
KW - Course recommender
KW - Exploration–exploitation
KW - Grade Point Average (GPA)
KW - Multi-armed bandits
KW - Thompson sampling algorithm
UR - https://www.scopus.com/pages/publications/105028324553
U2 - 10.1007/978-981-96-7140-3_40
DO - 10.1007/978-981-96-7140-3_40
M3 - Conference contribution
AN - SCOPUS:105028324553
SN - 9789819671397
T3 - Lecture Notes in Networks and Systems
SP - 595
EP - 607
BT - Advances in Data-Driven Computing and Intelligent Systems - Selected Papers from ADCIS 2024
A2 - Bansal, Jagdish Chand
A2 - Saha, Snehanshu
A2 - Coello, Carlos A.Coello.
A2 - Rathore, Hemant
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 20 September 2024 through 21 September 2024
ER -