Peningkatan Akurasi Metode K-Nearest Neighbor dengan Seleksi Fitur Symmetrical Uncertainty

Abstract

Accuracy of K-Nearest Neighbor (KNN) tends to be lower than other classification methods. The cause of this is related to the attributes used and the percentage of the influence of these attributes on the classification process in a data. And also attributes with less relevant influence can be a problem in determining the new class. One way that can be done to overcome this is by doing Feature Selection. In this research, the author selects features on K-Nearest Neighbor by using Symmetrical Uncertainty to remove attributes that have an unfavorable effect from the data set. Testing of the proposed method uses data sets obtained from the UCI Machine Learning Repository. The results obtained from testing the proposed method using feature selection with Symmetrical Uncertainty are able to increase the classification accuracy of KNN, with an increase in accuracy obtained after feature selection is 3.00 %.