Implementasi Algoritma FB-Prophet dan Algoritma ARIMA (Autoregressive Integrated Moving Average) dalam Memprediksi Harga Koin Cryptocurrency

Abstract

Cryptocurrency is a digital currency utilizing blockchain technology, with Bitcoin (BTC) and Ethereum (ETH) standing out as the most prominent examples. The potential for future price increases in cryptocurrencies, particularly BTC and ETH, has led people to consider them as attractive investment assets, seen as resistant to inflation for long-term storage. However, the highly volatile nature of the cryptocurrency market makes it challenging for individuals to conduct accurate technical analysis for predicting future prices. Many investors experience losses due to imprecise analyses. The aim of this research is to construct a website capable of predicting the daily, monthly, and yearly prices of BTC and ETH. This platform aims to provide a comprehensive overview of price movement trends, aiding investors and users in making informed decisions regarding cryptocurrency asset buying and selling, utilizing models with the highest accuracy. The research implements two prediction approaches for cryptocurrency prices, namely the FB-Prophet algorithm and the Autoregressive Integrated Moving Average (ARIMA) algorithm. The research utilizes MAPE, MSE, and RMSE metrics for model evaluation. The FB-Prophet model for BTC has a MAPE of 7.93%, MSE of 398,135,989.19, and RMSE of 19,953.35. For ETH, the MAPE of 8.13%, MSE of 388,269.76, and RMSE of 623.11. In contrast, the ARIMA model shows higher errors: for BTC, the MAPE of 22.54%, MSE of 16,140,020,334.26, and RMSE of 127,043.37, while for ETH, the MAPE of 23.68%, MSE of 29,811,728.88, and RMSE of 5,460.01. The evaluation results indicate that the FB-Prophet model is more effective in modeling and predicting cryptocurrency price fluctuations because the relatively low value indicates that the FB Prophet model has a small error rate.