PERBANDINGAN PENGARUH PENGGUNAAN EUCLIDIAN, MANHATTAN, DAN CHEBYCHEV TERHADAP TINGKAT AKURASI KLASIFIKASI

Abstract

The necessary for a classification system is needed to help facilitate human task. Classification is
to predict an object from a class that has no class. Many scientists are beginning to introduce
algorithms for classification, such as Artificial Neural Network, Support Vector Machine, KNearest

Neighbor,etc. This method is very simple, based on the closest distance between learning
data and objects. This distance is used as the value of similarity or closeness between the testing
data. Problems will appear we wrong to choose the metric of distance in a case that we will
classify.. The result is a decrease in the accuracy of our classification. In this study it can be
concluded that there are many alternative distance metrics that can be used in classification using
K-Nearest Neighbor algorithm. Changes in the use of standard distance metrics provide significant
changes in accuracy. From the above experiments also the authors can conclude that Manhattan's
distance metrics have better accuracy compared to other distance metrics in the case studies of
Breast Cancer.