K-Nearest Neighbor (K-NN) algorithm with Euclidean and Manhattan in classification of student graduation

Abstract

K-Nearest Neighbor (K-NN) algorithm is a classification algorithm that has been proven to solve various classification problems. Two approaches that can be used in this algorithm are K-NN with Euclidean and K-NN with Manhattan. The research aims to apply the K-NN algorithm with Euclidean and K-NN with Manhattan to classify the accuracy of graduation. Student graduation is determined by the variables of gender, major, number of first-semester credits, number of second-semester credits, number of third-semester credits, grade point on the first semester, grade point on the second semester, grade point on the third semester, and age. These variables determine the accuracy of student graduation, timely or untimely. The implementation of the K-NN algorithm is carried out using Rapidminer software. The results were obtained after testing 380 training data and 163 testing data.  The best accuracy system was achieved at K=7 with a value of 85.28%. The two algorithmic approaches did not affect the accuracy of the results. Furthermore, the addition of the value of K did not completely affect the accuracy.