A Comparative Study of Predicting Students’ Academic Performance Using Classification Algorithms

Depending on the different activities of the students, their target goals may change. There it is more important to identify the relation in their educational activities, other extracurricular activities, and behaviors. Therefore, classification algorithms are used to identify such patterns. It can be used to make predictions in education and it makes a huge contribution, especially in a competitive world. We examine what is the most appropriate classification algorithm for predicting students’ academic performance and those decisions based on information obtained from the graduates of the Sabaragamuwa University of Sri Lanka. The data set is preprocessed by cleaning and removing unwanted attributes and removing missing values etc. The preprocessed data is ranked using Waikato Environment for Knowledge Analysis (WEKA) data mining tool and to improve the results hyperparameter tuning was used. The classification algorithms such as Random Forest, Naïve Bayes, Multi-Layer Perceptron (MLP), and Decision Tree (J48) were used with the 80% training data and 20% testing dataset for creating a model. Random Forest algorithm presents best results considering that well-known statistical accuracy metrics such as accuracy, precision, recall, f-measure, Mean Absolute Error (MAE), and Root Mean Square Error (RMSE) values. Based on the Random Forest algorithm results, we can enhance the level of education and skills of students and create plans in the present to reach students’ future goals.