Clustering Kanker Serviks Berdasarkan Perbandingan Euclidean dan Manhattan Menggunakan Metode K-Means

Abstract

K-means a fairly simple and commonly used cluster of clusters to partition datasets into multiple clusters. Distance calculations are used to find similar data objects that lead to developing powerful algorithms for datamining such as classification and grouping. Some studies apply k-means algorithms using distance calculations such as Euclidean, Manhattan and Minkowski. The study used datasets from gynecological patients with a total of 401 patients examined and as many as 205 patients detected cervical cancer, while 196 other patients did not have cervical cancer. The results were shown with the help of confusion matrix and ROC curve, accuracy value obtained by 79.30% with ROC 79.17% on K-Means Euclidean Metric while K-Means Manhattan Metric by 67.83% with ROC 65.94%. Thus it can be concluded that the Euclidean method is the best method to be applied in the K-Means Clustering algorithm on cervical cancer datasets.