Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/157525
Title: Penerapan Algoritma Prophet untuk Peramalan Jangka Pendek Kurs Mata Uang Rupiah (IDR) ke Dolar Amerika Serikat (USD)
Other Titles: Application of Prophet Algorithm for Short-Term Forecasting of Exchange Rate of the Indonesian Rupiah (IDR) to United States of Dolar (USD)
Authors: Susetyo, Budi
Dito, Gerry Alfa
Sukma, Reza Arya
Issue Date: 2024
Publisher: IPB University
Abstract: Kurs mata uang yaitu ketidakstabilan dari harga mata uang yang disebabkan oleh penawaran dan permintaan mata uang suatu negara. Penggunaan kurs mata uang dalam menentukan posisi ekonomi suatu negara menjadi krusial mengingat dampak yang signifikan terhadap kebijakan moneter, perdagangan internasional, dan investasi. Salah satu metode yang dapat digunakan dalam peramalan adalah Algoritma Prophet. Prophet bekerja dengan meramalkan data deret waktu yang mana cocok dengan musiman, mingguan, dan harian, serta efek hari libur. Penelitian ini bertujuan untuk menentukan model prophet terbaik dan meramalkan kurs mata uang rupiah terhadap dolar Amerika Serikat dengan menggunakan model prophet terbaik. Penggunaan tahapan time series cross-validation dalam peramalan memberikan pengaruh yang signifikan saat penentuan model terbaik dengan nilai kombinasi hyperparameter yang berbeda. Model prophet terbaik adalah hasil hyperparameter tuning dengan changepoint prior scale 0,05, seasonality prior scale 1, dan seasonality mode multiplicative. Model tersebut cukup baik jika dilihat dari nilai RMSE menghasilkan nilai 45,3389. Hasil peramalan dengan model terbaik menunjukkan tren peningkatan nilai tukar rupiah terhadap dolar Amerika Serikat. Namun peramalan yang dilakukan oleh model kurang akurat dalam menangkap fluktuasi perubahan nilai tukar rupiah ke dolar Amerika Serikat.
Currency exchange rates are the instability of currency prices caused by the supply and demand of a country's currency. The utilisation of currency exchange rates in determining a country's economic position is crucial given its significant impact on monetary policy, international trade, and investment. One of the methods that can be used in forecasting is the Prophet Algorithm. Prophet works by forecasting time series data which is compatible with seasonal, weekly, and daily, as well as holiday effects. This thesis aims to determine the best prophet model and forecast the rupiah exchange rate against the United States dollar using the best prophet model. The usage of time series cross-validation stage in forecasting gives a significant effect when determining the best model with different hyperparameter combination values. The best prophet model is the result of hyperparameter tuning with changepoint prior scale 0,05, seasonality prior scale 1, and seasonality mode multiplicative. The model is quite good when viewed from the RMSE value resulting in a value of 45,3389. Forecasting results with the best model show an increasing trend in the rupiah exchange rate to the United States dollar. However, the forecasting performed by the model is less accurate in capturing fluctuations in the exchange rate changes of the rupiah to the United States dollar.
URI: http://repository.ipb.ac.id/handle/123456789/157525
Appears in Collections:UT - Statistics and Data Sciences

Files in This Item:
File Description SizeFormat 
cover_G1401201025_4fbdaf83370f4a5dbce664aa92da513a.pdfCover319.9 kBAdobe PDFView/Open
fulltext_G1401201025_ce5f0139f9264e679cb64330d65f7561.pdf
  Restricted Access
Fulltext1.66 MBAdobe PDFView/Open
lampiran_G1401201025_229efa8e73a14e3eabbe82909786bfb8.pdf
  Restricted Access
Lampiran96.86 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.