Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/159025
Title: Perbandingan K-Means dan K-Medoids Clustering dalam Pengelompokan Saham di Indonesia
Other Titles: Comparison of K-Means and K-Medoids Clustering in Grouping Stocks in Indonesia
Authors: Budiarti, Retno
Sumarno, Hadi
Puspita, Intan Dyah
Issue Date: 2024
Publisher: IPB University
Abstract: Pembentukan portofolio saham melibatkan keputusan investasi yang kompleks, terutama dalam memilih saham potensial di pasar modal. Metode clustering seperti K-Means dan K-Medoids dapat menyederhanakan proses ini dengan mengelompokkan saham berdasarkan karakteristik tertentu. Penelitian ini bertujuan menerapkan kedua metode tersebut untuk mengelompokkan 700 saham di Indonesia berdasarkan return, volatilitas, dan likuiditas. Hasil analisis menunjukkan K-Means membentuk tiga cluster dengan Within-Cluster Sum of Squares (WSS) terkecil pada cluster 3 sebesar 625,1118, sedangkan K-Medoids membentuk tiga cluster dengan WSS terkecil pada cluster 3 sebesar 709,5239. Evaluasi metode cluster menggunakan silhouette coefficient (SC) dan Dunn index (DI), menunjukkan K-Means lebih baik daripada K-Medoids dengan nilai SC dan DI terbesar. Profiling K-Means menunjukkan cluster 1 memiliki volatilitas dan return rendah, cluster 2 memiliki volatilitas dan likuiditas tinggi serta cluster 3 memiliki volatilitas rendah dan return tinggi. Hasil clustering ini diharapkan dapat membantu investor membentuk portofolio optimal berdasarkan karakteristik tertentu.
The formation of a stock portfolio involves complex investment decisions, especially in selecting potential stocks in the capital market. Clustering methods such as K-Means and K-Medoids can simplify this process by grouping stocks based on certain characteristics. This study aims to apply both methods to cluster 700 stocks in Indonesia based on return, volatility, and liquidity. The analysis results show K-Means forms three clusters with the smallest Within-Cluster Sum of Squares (WSS) in cluster 3 of 625.1118, while K-Medoids forms three clusters with the smallest WSS in cluster 3 of 709.5239. Evaluation of cluster methods using silhouette coefficient (SC) and Dunn index (DI), shows K-Means is better than K-Medoids with the largest SC and DI values. The K-Means profiling shows cluster 1 has low volatility and return, cluster 2 has high volatility and liquidity, and cluster 3 has low volatility and high return. The clustering results are expected to help investors form optimal portfolios based on certain characteristics.
URI: http://repository.ipb.ac.id/handle/123456789/159025
Appears in Collections:UT - Actuaria

Files in This Item:
File Description SizeFormat 
cover_G5402201015_7a21c95fde11459885f54dd1ecad6060.pdfCover1.41 MBAdobe PDFView/Open
fulltext_G5402201015_1cdee0b760a4473390a30c9b90fc93c3.pdf
  Restricted Access
Fulltext2.5 MBAdobe PDFView/Open
lampiran_G5402201015_732029edcee7446791d0d807b1cb7cdb.pdf
  Restricted Access
Lampiran669.5 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.