Dany Rahman 

Departemen Pendidikan Matematika, FPMIPA, Universitas Pendidikan Indonesia, Indonesia

Dewi Rachmatin 

Departemen Pendidikan Matematika, FPMIPA, Universitas Pendidikan Indonesia, Indonesia

Rini Marwati 

Departemen Pendidikan Matematika, FPMIPA, Universitas Pendidikan Indonesia, Indonesia

DOI: https://doi.org/10.19184/mims.v24i1.39159

ABSTRACT 

Cryptocurrencies are becoming one of the hottest topics in Indonesia's society. One of those issues concerns investors who incur financial losses as a result of investing in crypto. The facebook Prophet model, one of the forecast models, can offer a solution to this problem. The Prophet model is built using four function. These variables are trend, seasonality, holidays, and additional regressions. The Prophet model benefits from a number of advantages, one of which is its ability to generate decomposition graphs. The decomposition may give analysts more insight into the data they are analyzing. The Prophet model is used to forecast and decompose the price of a cryptocurrenciy called Solana in this study. A multiplicative model with linear function as trend function, weekly seasonality, and daily seasonality as seasonality function is the best model for Solana price forecasting and decomposition. Additionally, hyperparameters in the model are tuned so the model won’t suffer underfitting or overfitting indications. The fitted Prophet model is good at forecasting as a result of the evaluation process. As a result of the forecast and decomposition, the forecasted value and the decomposition graph of the Solana exchange rate for one hour later show that the price of Solana will remain constant.

Keywords: Cryptocurrency, time series, forecasting, decomposition, facebook prophet.

MSC2020: 62M10