TWITTER SENTIMENT ANALYSIS FOR BITCOIN PRICE PREDICTION

Abstract. In this paper, we have tried to provide a new way to predict the future price of Bitcoin by using Twitter messages and historical Bitcoin data. For this purpose, first we turn to machine learning models and use seven different models like Support Vector Machine (SVM), Na¨ıve Bayes, BERT (Bidirectional Encoder Representations from Transformers)+ Logistic Regression, BERT + Support Vector Machine (SVM), Neural Network, Ada boost classifier and Gradient boost classifier. In the next, we compare these models with each other. Finally, based on the results, we show that among these models, Gradient Boosting model has the highest accuracy.