Oktavian, Muhammad Diaz Syahmi (2026) PERFORMANCE ANALYSIS OF LSTM ALGORITHM IN HISTORICAL DATA BASED BITCOIN PRICE PREDICTION WITH MULTI PRECIOUS METAL PRICE COMPARISON VARIABLES. Undergraduate thesis, UPN VETERAN JAWA TIMUR.
|
Text (Cover)
cover fixx.pdf Download (765kB) | Preview |
|
|
Text (Bab 1)
CHAPTER I.pdf Download (212kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (BAB 2)
CHAPTER II.pdf Restricted to Repository staff only until 3 June 2029. Download (317kB) |
|
|
Text (BAB 3)
CHAPTER III.pdf Restricted to Repository staff only until 3 June 2029. Download (418kB) |
|
|
Text (BAB 4)
CHAPTER IV.pdf Restricted to Repository staff only until 3 June 2029. Download (623kB) |
|
|
Text (BAB 5)
CHAPTER V.pdf Download (158kB) | Preview |
|
|
Text (Daftar Pustaka)
REFERENCE.pdf Download (176kB) | Preview |
|
|
Text (Lampiran)
LAMPIRAN.pdf Restricted to Repository staff only until 3 June 2029. Download (4MB) |
Abstract
The extreme volatility of Bitcoin prices at high frequencies demands a highly responsive forecasting system. Traditionally, cryptocurrency fluctuations are solely associated with gold as a safe-haven asset; however, this approach often experiences a lagging phenomenon when responding to micro-market anomalies. This study aims to analyze the performance of the Long Short-Term Memory (LSTM) algorithm in predicting Bitcoin prices by comparing two architectures: a Bivariate scenario (utilizing Bitcoin and Gold variables) and a Multivariate scenario (integrating Multi-Precious Metal variables including Gold, Silver, Platinum, Palladium, and Rhodium). The evaluation was executed using time-series data with a 60-minute computational observation window. The experimental results indicated that the Bivariate model achieved an accuracy of 99.77% with a Mean Absolute Error (MAE) of 181.12 USD. In contrast, the Multivariate model demonstrated its superiority by increasing the accuracy to 99.84% and suppressing the MAE value to 130.24 USD (an error gap reduction of 50.88 USD). This reduction in computational error provides empirical confirmation that industrial metals (Platinum Group Metals) carry crucial predictive feature weights regarding cross-asset correlation. For operationalization, the algorithmic model was successfully implemented into a web-based Decision Support System using the Flask framework. This integrated system is capable of automating asynchronous API data fetching, facilitating interactive forecasting computation, and providing dynamic backtesting features to validate algorithmic transparency in real-time.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | H Social Sciences > HA Statistics H Social Sciences > HG Finance H Social Sciences > HG Finance > HG1709 Data processing H Social Sciences > HG Finance > HG4501-6051 Investment, capital formation, speculation Q Science > QA Mathematics > QA76.6 Computer Programming Q Science > QA Mathematics > QA76.87 Neural computers |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Muhammad Diaz Syahmi Oktavian | ||||||||||||
| Date Deposited: | 03 Jun 2026 08:04 | ||||||||||||
| Last Modified: | 03 Jun 2026 08:04 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/53525 |
Actions (login required)
 |
View Item |