Prediksi Harga Bitcoin Menggunakan Metode Long Short Term Memory

Investing has become a popular activity among the public as it can increase returns from such activities. Currently, there are various investment instruments available such as stocks, bonds, gold, property, and the latest is cryptocurrency. Since its inception in 2008, Bitcoin has emerged as the leading digital currency in terms of market capitalization and continues to attract investor attention. The first Bitcoin transaction occurred in January 2009. More than two years later, various reports estimated that Bitcoin circulation had exceeded 6,5 million with around 10000 users. This has led to a growing number of people interested in investing in Bitcoin. Based on these facts, it is important to predict Bitcoin prices to understand how Bitcoin prices will develop in the future. By predicting future Bitcoin prices, investors can estimate Bitcoin prices from the predicted prices in order to avoid making mistakes in investing and thus prevent significant potential losses. One way to predict Bitcoin prices is by using the Long Short-Term Memory (LSTM) method. One of the best models with a composition of 50 neurons and 400 epochs using Adam optimization, produce RMSE of 1027 and MAPE 1,76%, which mean the forecasting model has highly accurate forecasting. Then, to facilitate non-programmer users in forecasting, a web-based Graphical User Interface (GUI) is built using the Streamlit library.