This is the project code for the postgraduate dissertation (Data Science). The project focuses on the hourly Bitcoin price movements and returns prediction using the social media data, like Reddit posts, Twitter, and Google Trends. One of the aims of the study is to explore the relationship between public sentiment and Bitcoin price fluctuations. After that, we aim to compare the performance of different models, including machine learning models and deep learning models. Finally, the study leverages the optimal model and features to build the action recommendation model to maximize profits in the short and long term.
Sentiment Analysis Tools: Vader, Textblob, Pysentiment2
Dataset:
Reddit Posts: https://www.kaggle.com/datasets/leukipp/reddit-crypto-data Tweets: https://www.kaggle.com/datasets/hiraddolatzadeh/bitcoin-tweets-2021-2022 Google Trends: Pytrends API
Models: LSTM, CNN, CNN-LSTM, ARIMA, Random Forest Classifier