Optimization Of Determination Against K-Means Cluster Algorithm Using Elbow Creation

Abstract

Clustering is a data mining method for grouping data that have similar or different characters in each section. One of the methods is using K-Means by measuring the distance between clusters using the shortest distance or Euclidean Distance. K-means entails weakness, which is the determination of clusters in k-means clustering, resulting in the different data grouping and affecting the results of the data cluster distribution. To overcome this issue, the elbow creation method is employed to determine the similarity level in the cluster by observing the comparison between Root Means Square and R Square to measure the homogeneity and heterogeneity of the cluster where this method is applied by considering the changes in the comparison between the RMSSTD (Root Means Square Standard Deviation) and RS (R Squared) values which have the intersection of the RMSSTD and RSquared values. The difference between RMSSTD cluster 1 and RMSSTD cluster 2 was 0.066 and RS cluster 1 and RS cluster 2 was -0.304. Based on those figures, the highest difference was found in cluster 2. All considered, tourist destinations in East Asia frequently visited or interested to visitors are grouped into cluster 2, comprising criteria 6, 7, 8, and 10, or in other words, resort destination, picnic area, beaches, and religious institutions