SCSYGPS: Students Course-Wise, Semester-Wise and Year-Wise Grade Prediction System in Education Sector Using Supervised Learning

Educational Data Mining (EDM) is the research field that deals with the application of Data Mining techniques on the data obtained from education sector. Predicting students' academic performance is of paramount importance to students, instructor, and educational institutes. In this research article, a framework for prediction of students' performance in particular course, semester and year is developed and named it as S
CSY
GPS (Students' Course-wise, Semester-wise, and Year-wise Grade Prediction System). In this framework, data is collected from university result ledgers related to four year Computer Science and Engineering Programme and prepared nine datasets. Eight classification algorithms - Random Forests (RF), Linear Support Vector Classifier (LSVC), Gradient Boosting (GB), K-Nearest Neighbour (KNN), AdaBoost, XGBoost, LightGBM, and Linear Discriminannt Analysis (LDA) are applied one by one on nine datasets. The classification algorithm that works best for each dataset is selected based on performance parameters values such as Accuracy, F1-Score, Precision, Recall, and Mean Absolute Error (MAE). The classification algorithm which has the highest value of performance parameters accuracy, f1-score, precision, recall, and lowest value of MAE is considered as the best classification algorithm for that dataset. For seven datasets, LightGBM is found to be the best classification algorithm while XGBoost and AdaBoost is noted to be the best classification algorithms for two datasets with accuracy 97% nd 95% respectively.