|
Original research EXPLORING THE IMPACT OF SOCIAL MEDIA SENTIMENT ON BITCOIN PRICE MOVEMENTS: A TIME SERIES FORECASTING APPROACH USING LSTM MODELSPages 155-164 Abstract
The financial industry underwent a major change after Bitcoin creation together with other cryptocurrencies established Bitcoin as the dominant digital asset. The price fluctuations of Bitcoin affected by market demand and technological changes alongside regulatory factors and overall economic conditions together with social media buzz create both profitable and dangerous prospects for financial investors. Multiple social media sites including Twitter and Reddit together with Facebook serve as leading indicators for Bitcoin price movements. This study analyzes Bitcoin price movements in relationship to social media opinion through the application of machine learning (ML) and deep learning (DL) models to boost forecasting precision. The Long Short-Term Memory (LSTM) network serves as the main predictive model to forecast Bitcoin's price through the combination of historical price information with sentiment data obtained from social media postings. The LSTM network excelled over Random Forest, XGBoost, and RNN because it produced MAE 0.001562 together with MSE 0.000147 and RMSE 0.012119 and R² 0.992604 score. Random Forest exhibited MAE of 7.07822 alongside R² of 0.999999 but XGBoost delivered an MAE of 0.00063 together with R² of 0.99983. The research demonstrates how sentiment analysis can function as a powerful tool for predicting cryptocurrency market price changes thus delivering important market intelligence to traders and investors.
Keywords: Bitcoin, Cryptocurrency, Sentiment Analysis, Machine Learning, Deep Learning, LSTM, Social Media, Price Forecasting, Financial Modeling, XGBoost, Random Forest, RNN.
|
Thakit T. Yousif 
PDF 