Analysis on relationship between bitcoin price trend and sentiment of bitcoin related tweets by ML and NLP
Konferenz: ICMLCA 2021 - 2nd International Conference on Machine Learning and Computer Application
17.12.2021 - 19.12.2021 in Shenyang, China
Tagungsband: ICMLCA 2021
Seiten: 5Sprache: EnglischTyp: PDF
Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt
Autoren:
Jiang, Nan (Union College, Schenectady, New York, USA)
Ti, Chang (Drew University, Madison, New Jersey, USA)
Mao, Yuxin (Shanghai World Foreign Language Academy, Shanghai, China)
Wu, Tianhao (Changjun High School International Department, Changsha, Hunan, China)
Inhalt:
It is noticed that sentiment of related tweets has great influence on the price trend of cryptocurrency, especially bitcoin. This research focuses on bitcoin-related tweets and bitcoin price in January 2019 to May 2019, and employs Natural Language Process (NLP) and Machine Learning (ML) to study the relationship between them. In this work, tweets are analyzed by vaderSentiment and TextBlob respectively. The best trained model is Naive Bayes with TextBlob as sentiment analysis. It reaches 0.729 for F1 score and 0.641 for precision. This research makes a contribution to 1) comparing vaderSentiment and TextBlob on tweets sentiment analysis and 2) training models to learn relationship between bitcoin related tweets and bitcoin price trends in the following days.