Predicting the Impact of Articles on Bitcoin’s Price Change
The burgeoned interest in cryptocurrency in recent years has led to a rise in investment of bitcoin. Ever since its release in 2009, bitcoin’s price has been fluctuating and investors have spent lots of effort speculating and predicting the change in bitcoin’s price. Our team finds it extremely fruitful to be able to provide insight into the influence of online news articles on bitcoin price. Coupled with the chatbot, it can serve as a platform to provide investors with information with which to make a trade or buy decision in the bitcoin market.
Our team has utilized both supervised machine learning and semi-supervised machine learning methods to label an article and predict whether it will positively or negatively impact bitcoin price. The supervised learning model is based on unique features of an article, such as its author and publisher, the sentiment and magnitude of the article and the term frequency–inverse document frequency score of an article. The semi-supervised learning model utilizes a subset of already labeled articles to try and predict the label of unlabeled articles, and can be used in the future to provide a rank of high-impact articles.
https://github.com/jeffpengjeff/Paradigm_Team_1_Fall18.git