Best Student Classification using Ensemble Random Forest Method

Abstract

Education at Madrasah Aliyah Negeri 1 Medan considers religious values, ethics, leadership, and participation in extracurricular activities as an integral part of student character building. Therefore, it is necessary to develop a classification system that integrates these various aspects to ensure that the best students not only excel in academic exams but also have strong social, leadership, moral and extracurricular abilities. The purpose of this research is to implement the Random Forest Ensemble method in predicting the best students of MAN 1 Medan and build a system in predicting the best students of MAN 1 Medan using the Random Forest Ensemble method. The data used is 550 divided into 385 data as Training data and 165 Testing data. In the implementation of Random Forest with three decision trees formed from entropy calculations on 385 training data, followed by testing using 10 testing data from a total of 165 existing data, the results show that the model predicts 8 data as class 1 (best students) and 2 data as class 0 (normal students) from a total of 10 testing data. From the test results using 385 training data and 165 testing data, the Random Forest model predicted 70 data as the best students (class 1) and 95 data as normal students (class 0) with high precision for both classes (0.94 for class 0 and 0.99 for class 1), as well as high recall for both classes (0.92 for class 0 and 0.99 for class 1) The overall accuracy reached 0.96, confirming the model's ability to classify the data well overall.