Evaluation of the implementation of machine learning algorithm K-Nearest Neighbors (KNN) using rapid miner on junior high school student learning outcomes

Abstract

The focus of the PISA 2022 assessment is on the subjects of Mathematics, Language and Science. Therefore, these subjects are compulsory subjects at every level of education. Because learning activities are the most important activities, the success of a learning activity is measured by the learning outcomes that have reached completeness or failed. Prediction of completeness or failure can be done by classifying data using the K-Nearest Neighbors (KNN) algorithm using the RapidMiner application. The KNN algorithm is one of the classification methods for a set of data based on learning data that has been classified before. The data used are student learning outcomes in Mathematics, Indonesian Language and Science subjects at the junior high school education level in Padang city. This research aims to predict student learning outcomes in Mathematics, Indonesian and Science subjects based on student score completeness by comparing various k values to obtain the best performance of this algorithm. The results obtained after analyzing the KNN algorithm are Classification using the KNN algorithm is most accurate when the value of k = 5 and k = 7. Where by using the value of k, the accuracy of the KNN algorithm reaches the maximum result of 94.12%. Thus, this algorithm can help teachers to predict or find out how appropriate student completeness.