With this year’s bitcoin boom, cryptocurrency has been regularly making headlines. Cryptocurrencies fluctuate daily by large amounts. Perhaps most notably, some currencies experience massive rises or falls at times, appreciating even by 1000% in a day. Given the large volatility, it is possible that part of the reason why price changes is due to user sentiment. Twitter is perhaps one of the best available websites to use for sentiment prediction, as it has a large userbase and accesible API. This project aims to predict price changes through tweet sentiment analysis, through a multivariate time series model.
The project Jupyter Notebook can be viewed by clicking this link.