CD Tesis

Algoritma Dbk-Means Clustering Berbasis Feature Engineering Dalam Pembentukan Portofolio Saham Berdasarkan Model Mean-Var

---

ABSTRACT
RATNA TRI AULIA NIM. 2210246886, DBK-Means Clustering Algorithm
Based on Feature Engineering in Stock Portfolio Formation Using the
Mean-VaR Model, Supervised by Arisman Adnan dan Ihda Hasbiyati.
Stock investment offers high profit potential but also involves unavoidable
risks, thus requiring appropriate strategies to determine the right combination of
investment assets. The DBK-Means clustering algorithm based on feature engineering
serves as a promising alternative for stock selection, combined with the
Mean-Value at Risk (Mean-VaR) approach to construct an optimal portfolio that
balances return and risk. This algorithm is applied to group stocks included in the
LQ45 index, which consists of the most liquid and best-performing stocks on the
Indonesia Stock Exchange. The data used consist of daily closing prices of LQ45
stocks from the 2020–2024 period, taking into account market dynamics between
the pandemic and post-pandemic periods through the Independent Sample t-test.
The feature engineering process involves constructing three key indicators volatility,
liquidity, and market capitalization, which are then normalized before the
clustering process. This study aims to determine the number of clusters formed,
identify investment-worthy stocks, and calculate the allocation weights and risk
levels of the optimal portfolio. The results show that seven clusters were formed
with good grouping structure, as reflected by a Silhouette Coefficient (SC) of
0.7322. From each cluster, the stock with the highest expected return was selected
namely ASII, BMRI, BBCA, ICBP, PTBA, INCO, and KLBF, which were
then used for portfolio construction. Further analysis indicates that this portfolio
is capable of producing optimal asset combinations within a risk tolerance range
of 1.5 ≤ τ ≤ 4.2 without employing a short-selling strategy. All analyses were
performed using RStudio version 4.4.2. The use of the DBK-Means clustering
algorithm and the Mean-VaR approach has proven effective in filtering potential
stocks and constructing a well-balanced portfolio in terms of return and risk.
Keywords: DBK-Means Clustering, Feature engineering, Mean-VaR, Silhouette
coefficient

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

2210246886

Penerbit

Pekanbaru. : Universitas Riau – PASCA– Magister Matematika., 2025

Deskripsi Fisik

-

Bahasa

Indonesia

ISBN/ISSN

-

Klasifikasi

2210246886

Tipe Isi

-

Tipe Media

-

Tipe Pembawa

-

Edisi

-

Subjek

MAGISTER MATEMATIKA

Info Detail Spesifik

-

Pernyataan Tanggungjawab

RIDHO

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar