PREDIKSI HARGA MATA UANG BITCOIN MENGGUNAKAN METODE OPTIMIZED GATED RECURRENT UNIT DENGAN OPTIMASI ADAPTIVE MOMENT ESTIMATION

Nugroho, Wahid Rochman (2022) PREDIKSI HARGA MATA UANG BITCOIN MENGGUNAKAN METODE OPTIMIZED GATED RECURRENT UNIT DENGAN OPTIMASI ADAPTIVE MOMENT ESTIMATION. Other thesis, UPN "Veteran" Yogyakarta.

Preview	Text ABSTRAK.pdf Download (403kB) | Preview
Preview	Text DAFTAR PUSTAKA.pdf Download (533kB) | Preview
Preview	Other HALAMAN PENGESAHAN PEMBIMBING.PDF Download (200kB) | Preview
Preview	Text DAFTAR PUSTAKA.pdf Download (533kB) | Preview
Preview	Text COVER.pdf Download (243kB) | Preview
	Text SKRIPSI-FULL-Wahid-Rochman-Nugroho-123160044.pdf Restricted to Repository staff only Download (1MB)

Abstract

vii
ABSTRAK
Bitcoin pertama kali diciptakan oleh Satoshi Nakamoto pada tahun 2008 dan
mengudara pada tahun 2009. Bitcoin merupakan hasil penelitian Satoshi Nakamoto tentang
Sistem Pembayaran Peer to Peer yang menggunakan jaringan Blockchain dan konsensus .
Semenjak kemunculannya Bitcoin menjadi populer karena sistem yang ter-desentralisasi dan
tidak membutuhkan pihak institusi ketiga untuk memverifikasi pembayaran, sedangkan saat
ini Bitcoin sangat populer di kalangan pebisnis dan investor. Hal tersebut dipicu karena
kenaikan signifikan nilai mata uang Bitcoin yang membuat para investor dan pebisnis
tertarik untuk berinvestasi dan mendapatkan keuntungan. Akan tetapi nilai mata uang
Bitcoin tidaklah selalu bertambah setiap waktu, ada kalanya harga mata uang Bitcoin turun
dan naik secara signifikan dengan sangat cepat sehingga dapat menimbulkan kerugian. Oleh
karena itu dibutuhkan suatu prediksi untuk memprediksi harga mata uang Bitcoin.
Tahap awal penelitian ini adalah pengumpulan data dari sumber yang terpercaya dan
menganalisis data yang telah dikumpulkan. Lalu dilanjutkan dengan pengolahan data dengan
tahap preprocessing. Pada tahap preprocessing dilakukan pembersihan data, normalisasi
dengan Min-Max dan proses Sliding Window. Setelah pembersihan data, proses normalisasi
dilakukan untuk menjadikan data sesuai dengan skala 0 sampai 1. Lalu dilanjutkan dengan
proses Sliding Window untuk membagi nilai menjadi X dan Y. Setelah tahap preprocessing
dilakukan maka akan dilakukan pelatihan dengan menggunakan metode Optimized Gated
Recurrent Unit (OGRU) dengan optimasi Adaptive Moment Estimation (ADAM).
Pada penelitian ini dilakukan pelatihan OGRU-ADAM dengan data historical
Bitcoin, data Google Trend dan data harga emas untuk memprediksi harga Bitcoin.
Penelitian dilakukan dengan mengukur akurasi hasil prediksi dan learning efficiency. Hasil
pada penelitian menghasilkan bahwa akurasi prediksi terbaik didapatkan dari model OGRUADAM dengan akurasi Mean Absolute Percentage Error (MAPE) 0.597895834 dan Mean
Squared Error (MSE) 4.707695E-05. Sedangkan hasil learning efficiency didapatkan hasil
terbaik dari model OGRU-ADAM dengan nilai MAPE 2.7651236560 dan MSE
0.0004268393.
Kata kunci : bitcoin, prediksi, google trend, emas. Gold, optimized gated recurrent unit,
recurrent neural network, adam optimizer, adaptive moment estimation, forecast, prediksi.viii
ABSTRACT
Bitcoin was first created by Satoshi Nakamoto in 2008 and went live in 2009. Bitcoin
is the result of Satoshi Nakamoto's research on Peer to Peer Payment Systems that use
Blockchain and consensus networks. Since its inception Bitcoin has become popular because
the system is decentralized and does not require a third party institution to verify payments,
while Bitcoin is currently very popular among business people and investors. This was
triggered by the significant increase in the value of the Bitcoin currency, which attracted
investors and business people to invest and earn profits. However, the value of the Bitcoin
currency does not always increase all the time, there are times when the price of the Bitcoin
currency drops and rises significantly very quickly so that it can cause losses. Therefore, a
prediction is needed to predict the price of the Bitcoin currency.
The initial stage of this research is collecting data from reliable sources and analyzing
the data that has been collected. Then proceed with data processing with the preprocessing
stage. In the preprocessing stage, data cleaning, normalization with Min-Max and Sliding
Window processes are carried out. After cleaning the data, the normalization process is
carried out to make the data according to a scale of 0 to 1. Then proceed with the Sliding
Window process to divide the values into X and Y. After the preprocessing stage is carried
out, training will be carried out using the Optimized Gated Recurrent Unit (OGRU) method
with optimization of Adaptive Moment Estimation (ADAM).
In this study, OGRU-ADAM training was conducted with historical Bitcoin data,
Google Trend data and gold price data to predict Bitcoin prices. The research was conducted
by measuring the accuracy of the prediction results and learning efficiency. The results of
the study showed that the best prediction accuracy was obtained from the OGRU-ADAM
model with an accuracy of Mean Absolute Percentage Error (MAPE) 0.597895834 and
Mean Squared Error (MSE) 4.707695E-05. While the results of learning efficiency obtained
the best results from the OGRU-ADAM model with a MAPE value of 2.7651236560 and an
MSE of 0.0004268393.
Keywords: bitcoin, prediksi, google trend, emas. Gold, optimized gated recurrent unit,
recurrent neural network, adam optimizer, adaptive moment estimation, forecast, prediksi.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	bitcoin, prediksi, google trend, emas. Gold, optimized gated recurrent unit, recurrent neural network, adam optimizer, adaptive moment estimation, forecast, prediksi
Subjects:	Z Bibliography. Library Science. Information Resources > ZA Information resources
Divisions:	Faculty of Engineering, Science and Mathematics > School of Engineering Sciences
Depositing User:	Eko Yuli
Date Deposited:	15 Feb 2022 03:49
Last Modified:	27 Oct 2022 06:44
URI:	http://eprints.upnyk.ac.id/id/eprint/28377

Actions (login required)

View Item