Intelligent System for Student Performance Prediction Using Machine Learning

Abstract

Accurately predicting student performance remains a significant challenge in the educational sector. Identifying students who need additional support early can significantly impact their academic outcomes. This study aims to develop an intelligent solution for predicting student performance using supervised machine learning algorithms. This proposed focus on addressing the limitations of existing prediction models and enhancing prediction accuracy. In this work employed three supervised machine learning algorithms: Random Forest, Extra Trees, and K-Nearest Neighbors. The steps of research methodology contained (data collection, preprocessing, feature identification, model construction, and evaluation). This paper utilized a dataset comprising 24,000 training instances and 6,000 testing instances, applying various preprocessing techniques for data optimization. The Extra Trees algorithm achieved the highest accuracy (98.15%), followed by Random Forest (94.03%) and K-Nearest Neighbors (91.65%). All algorithms demonstrated high precision and recall. Notably, K-Nearest Neighbors exhibited exceptional computational efficiency with a training time of 0.00 seconds. This study proposed an efficient model for prediction student performance. The high accuracy and efficiency of the proposed system highlight its potential for application in educational data mining. The findings of this proposed to improving student success rates in educational institutions by enabling timely and appropriate interventions.

Keywords

• Artificial Intelligence
• Student Performance Prediction
• Supervised Machine Learning
• Educational Data Mining
• Extra Trees Algorithm