Student performance prediction based on machine learning algorithms at the Adventist University of Central Africa

Abstract

In recent years, there has been a growing interest in predicting student performance in educational institutions using machine learning. This paper presents an approach to predict student performance based on data mining classification techniques. The objective was to develop a predictive model that can accurately forecast student performance and identify factors that influence academic success. The proposed approach involves the collection of various attributes related to students, such as identification numbers, department, faculty, GPA earned, nationality, place of birth, religion, gender, student absence days, parents' school satisfaction, and principal pass. These attributes are used as input variables for classification algorithms, including decision trees, logistic regression, and support vector machines. The dataset used in this study is obtained from University in the office of registrar and consists of historical data of students, including their grades and other relevant information. The dataset is preprocessed to handle missing values, outliers, and categorical variables. The dataset has 25 features and 2583 observations which are enough for training our models in doing experiment. Different classification algorithms are applied to the preprocessed dataset, and their performance is evaluated using various evaluation metrics, such as accuracy, precision. The experimental results indicate that the proposed approach achieves promising performance in predicting student performance. The decision tree algorithm outperforms other classifiers with the highest accuracy. The selected features reveal that factors such as previous academic performance, identification numbers, department, faculty, GPA earned. Influence student performance. Findings showed that the decision tree predicts student performance with accuracy of 99.87%. By leveraging student-related attributes and employing machine learning algorithms, educational institutions can enhance their decision-making processes and provide targeted support to students, ultimately improving overall educational outcomes.