A Deep Learning Framework for Hourly Bitcoin Price Prediction Using Bi-LSTM and Sentiment Analysis of Twitter Data

Autor: Patel, Raj, Chauhan, Jaya, Tiwari, Naveen Kumar, Upaddhyay, Vipin, Bajpai, Abhishek
Zdroj: SN Computer Science; August 2024, Vol. 5 Issue: 6
Abstrakt: Cryptocurrencies like bitcoin predictions hold significant importance due to their impact on various stakeholders and aspects of the financial landscape. Investors heavily rely on the predictions to make effective and efficient conclusions about buying, selling, or holding Bitcoin as part of their investment portfolios. Accurate forecasts enable effective risk management, allowing individuals and institutions to confidently navigate the volatile cryptocurrency market. In this paradigm, our emphasis is on the prediction of the next hour’s bitcoin price. In most of the model, we give the bitcoin price as input but here along with the bitcoin price we are giving coin price and sentiments as input and it predicts the next hour’s output price. In this work, we have used bidirectional LSTM with preprocessed data. We have performed data cleansing and sentiment analysis as part of the data pretreatment step. Recent 90-days data have been selected for the prediction which is then fed to a bidirectional LSTM model. It is a specialized version of recurrent neural network (RNN) architecture that is often used in sequence-to-sequence tasks, including time series prediction. It can be applied to Bitcoin price prediction by leveraging its ability to capture dependencies in both past and future data. The source code of the project is available at: https://github.com/rajpatel4835/Forecasting-Bitcoin-Price.
Databáze: Supplemental Index
Externí odkaz: