Non-Hierarchical Cluster Analysis to Group Tektite in Australian Strewnfield Based on Composition of Geochemical Data Using the K-Means Clustering Method
Authors: Juli Mulyanto (a), Triyana Muliawati (a*)

a) Department of Mathematics, Institut Teknologi Sumatera
Jalan Terusan Ryacudu, Lampung 35365, Indonesia
*triyana.muliawati[at]ma.itera.ac.id

Abstract

Cluster analysis aims to group objects based on the similarity level of characteristics among these objects. This study examined cluster analysis to classify tektites in the Australian Strewnfield based on the composition of geochemical data using the k-means clustering method. The data used were compiled from as much as 60 tektite samples, and the variables used were nine geochemical data components consisting of {SiO}_2,\ {\ TiO}_2,\ {\ Al}_2O_3,\ \ FeO,\ M\ nO,\ \ MgO,\ \ CaO,\ {\ Na}_2O,\ and K_2O. Before grouping using the k-means clustering method, principal component analysis is carried out to reduce dimensional data that can represent diversity and the Silhoutte method to obtain optimal k cluster values. Based on the results of the main component analysis, 2 main components were formed from the 9 variables analyzed. And based on the results of the grouping analysis using the k-means clustering method, the results show that the tektite in the Australian Strewnfield is divided into three clusters with a silhouette value of 0.6068 , that significanly depend on silica minerals abundances ({SiO}_2).

Keywords: Tektite- Silhouette Method- Principal Component Analysis- Geochemistry- K-Means Clustering

Topic: MATHEMATICS AND STATISTICS