Skripsi

PERBANDINGAN PERFORMA LINEAR REGRESSION DAN XGBOOST UNTUK PREDIKSI HARGA BITCOIN BERDASARKAN INDIKATOR TEKNIKAL

---

Penilaian

---

This research aims to compare the performance of the Linear Regression model with various feature selection methods and the XGBoost model with hyperparameter tuning in predicting Bitcoin prices. The Bitcoin price prediction model is very important for traders and investors to understand price movement patterns and make better investment decisions. The data used is historical Bitcoin price data processed by adding technical indicators such as Moving Average, RSI, MACD, Bollinger Bands, Momentum, Stochastic Oscillator, and Rate of Change (ROC). The Linear Regression method is implemented with five feature selection approaches, namely without feature selection, M5 Prime, Greedy Selection, T-Test, and Iterative T-Test. While the XGBoost model is optimized through systematic tuning of the max_depth and n_estimators parameters. Evaluation of both models is carried out using the Root Mean Square Error (RMSE) and Relative RMSE metrics with a 5-fold cross validation technique to ensure the validity of the results. The results of the study show that Linear Regression with the Greedy Selection method provides the best performance with an RMSE of 267.39, compared to XGBoost with optimal parameters (max_depth=8, n_estimators=900) which produces an RMSE of 443.7334. Feature importance analysis reveals that Linear Regression uses 12 selected features, while XGBoost tends to focus on the basic price features (Low and High) by ignoring several technical indicators such as RSI and BB_Upper which have importance close to zero. This finding indicates that for the dataset used, the Linear Regression model with proper feature selection can produce more accurate predictions than more complex models such as XGBoost.

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2507006354	T184499	T1844992025	Central Library (Reference)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1844992025

Publisher

Indralaya : Prodi Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2025

Collation

xv, 109 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

338.507

Content Type

Text

Media Type

unmediated

Carrier Type

other (computer)

Edition

-

Subject(s)

Teori Harga
Prodi Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment