BITCOIN PRICE PREDICTION USING DEEP LEARNING

Abstract

Bitcoin is a sort of computerized cash in which a record of exchanges is kept up and new units of cash are produced by the computational arrangement of scientific issues, and which works autonomously of a central bank. The price of a bitcoin is equal to 7000 USD and is growing more day by day. Since this is not a low price, approximately 10 million people are investing on it. This would be exceptionally curiously for investors to figure the Bitcoin esteem but the same time making it troublesome to predict. This study centers on predicting daily Bitcoin prices using some models (MLP, LSTM, CNN, GRU, ARIMA) of deep learning. In each of these models we will interpret the train and test loss, mean squared error, the performance in time and in the end predicted prices. After interpreting all this, we will compare all these results between different models.