Application of the K-Nearest Neighbor Algorithm to Analyze the Learning Ability of Grade VII Students in English Subjects

Abstract

SMP Negeri 2 Pahungan Lodu currently lacks a system for classifying students based on their English language learning abilities. As a result, the learning process remains generalized, failing to account for the varying levels of student understanding. This situation poses challenges for teachers in adapting their instructional methods, leading to suboptimal academic interventions for students experiencing learning difficulties. To address these challenges, this study employs the K-Nearest Neighbor (KNN) algorithm as a classification method to categorize students into three groups: Able, Quite Able, and Underable. These categories are determined based on academic data, including assignment scores, practice assessments, midterm (UTS) and final exam (UAS) scores, report card grades, and student attendance levels. This research utilizes a data mining approach with the KNN algorithm, which operates by calculating the Euclidean distance between student data points and assigning categories based on the nearest neighbors. The dataset used in this study comprised 63 students after undergoing data cleaning. Subsequently, the data was divided into 50 training samples and 13 test samples. The results indicate that the KNN algorithm successfully classifies the test data with an accuracy rate of 84.62%. These findings demonstrate that the KNN algorithm is an effective tool for academic decision-making and for developing learning strategies tailored to students' ability levels.