Investigate a connection between mass-media sentiment of Bitcoin and its value over time (regardless of whether that connection is based on causation or correlation).
- Uses NewsAPI to fetch all articles from the past 24 hours that mention Bitcoin
- Adds up the number of positive and negative words per article title from modified word lists compiled by Minqing Hu and Bing Liu from the University of Illinois at Chicago (see citations 1 and 2) to get a net for that article (a positive number or negative number)
- Finds the percentage/ratio of sentiment that day
- Gets Bitcoin's closing price from the previous day
- Find the percentage agreement between price and sentiment to determine if a connection exists
- Exports this data to a CSV
- Python 3.x
- NewsApi api key
- NewsAPI python wrapper
- Minqing Hu and Bing Liu. "Mining and Summarizing Customer Reviews." Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD-2004), Aug 22-25, 2004, Seattle, Washington, USA.
- Bing Liu, Minqing Hu and Junsheng Cheng. "Opinion Observer: Analyzing and Comparing Opinions on the Web." Proceedings of the 14th International World Wide Web conference (WWW-2005), May 10-14, 2005, Chiba, Japan.