Identifikasi Pola Perilaku Pasar Cryptocurrency Menggunakan Metode Clustering Berbasis Data Harga dan Volume Perdagangan

Hadikusuma, Agnes Priscilla Sekartaji (2026) Identifikasi Pola Perilaku Pasar Cryptocurrency Menggunakan Metode Clustering Berbasis Data Harga dan Volume Perdagangan. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025221295-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (11MB) | Request a copy

Abstract

Pasar cryptocurrency memiliki sifat volatilitas yang tinggi serta harga dan volume perdagangan yang kompleks, sehingga menyulitkan identifikasi pola perilaku pasar cryptocurrency. Penelitian ini mengidentifikasi pola perilaku pasar cryptocurrency menggunakan metode clustering berbasis harga dan volume perdagangan melalui unsupervised learning. Data yang digunakan dari empat aset cryptocurrency, yaitu Bitcoin (BTC), Ethereum (ETH), Solana (SOL), dan XRP, dalam periode Januari 2020 sampai dengan Desember 2024 yang diambil dari Yahoo Finance. Data penelitian ini telah melewati tahap preprocessing yang meliputi data cleaning, feature engineering, feature selection, dan normalisasi. Proses feature engineering dilakukan dengan menurunkan fitur harga dan volume perdagangan, seperti return, volatilitas, harga, dan volume perdagangan. Penelitian ini menggunakan metode clustering dengan algoritma K-Means, Density-Based Spatial Clustering of Applications with Noise (DBSCAN), Gaussian Mixture Model (GMM), dan Spectral. Evaluasi model menggunakan metrik internal Silhouette Score, Davies-Bouldin Index (DBI), dan Calinski-Harabasz Index (CHI), serta metrik eksternal Adjusted Rand Index (ARI) dan Normalized Mutual Information (NMI), sedangkan visualisasi menggunakan t-Distributed Stochastic Neighbor Embedding (tSNE). Hasil penelitian menunjukkan bahwa algoritma Spectral menghasilkan nilai ARI sebesar 0,2179 dan NMI sebesar 0,2592 yang merupakan nilai tertinggi di antara keempat algoritma, dengan distribusi fase Bearish sebesar 17,74%, fase Sideways sebesar 62,52%, dan fase Bullish sebesar 19,74% yang paling mendekati Ground Truth.
==================================================================================================================================
The cryptocurrency market is characterized by high volatility and complex price and trading volume dynamics, making it challenging to identify market behavior patterns. This study aims to identify cryptocurrency market behavior patterns using clustering methods based on price and trading volume through an unsupervised learning approach. The dataset consists of four cryptocurrency assets Bitcoin (BTC), Ethereum (ETH), Solana (SOL), and XRP covering the period from January 2020 to December 2024, sourced from Yahoo Finance. The data underwent preprocessing stages, including data cleaning, feature engineering, feature selection, and normalization. The feature engineering process involved deriving features from price and trading volume, such as returns, volatility, price-based ratios, and volume-based indicators. This study applies clustering methods using four algorithms K-Means, DensityBased Spatial Clustering of Applications with Noise (DBSCAN), Gaussian Mixture Model (GMM), and Spectral. Model evaluation using internal metrics, Silhouette-Score, DaviesBouldin Index (DBI), and Calinski- Harabasz Index (CHI), as well as external metrics Adjusted Rand Index (ARI) and Normalized Mutual Information (NMI), Additionally, visualization is performed using t-Distributed Stochastic Neighbor Embedding (t-SNE). The results indicate that Spectral achieves the best performance, with an ARI value of 0,2179 and an NMI value of 0,2592, outperforming the other algorithms. The resulting distribution Bearish phase 17,74%, Sideways phase 62,52%, and Bullish phase 19,74% is the closest to the Ground Truth, indicating its superior capability in capturing meaningful market behavior patterns.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	clustering, cryptocurrency, dbscan, gaussian mixture model, harga, k-means, perilaku pasar, spectral, unsupervised learning, volume. ========================================================================================================================== clustering, cryptocurrency, dbscan, gaussian mixture model, k-means, market behavior, price, spectral, unsupervised learning, volume
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Agnes Priscilla Sekartaji Hadikusuma
Date Deposited:	25 Jun 2026 00:47
Last Modified:	25 Jun 2026 00:47
URI:	http://repository.its.ac.id/id/eprint/134037

Actions (login required)

View Item