This project combines Bitcoin price data and news from various crypto websites to create a comprehensive database for analysis. The main goals are to summarize news articles, perform sentiment analysis, and explore correlations between Bitcoin prices and crypto news sentiments.
- Data Collection:
- Downloaded Bitcoin price data from Binance.
- Built scrapers to collect news from multiple crypto websites.
- Data Integration:
- Combined Bitcoin price data and news articles into a unified database.
- News Summarization:
- Used the distilbart-cnn-12-6 model for summarizing large volumes of news.
- Subabase and Azure Cloud for efficient summarization.
- Sentiment Analysis:
- Analyzed summarized news using four models:
- Analysis:
- Evaluated model performance using dispersion metrics and correlation matrices.
- Final Output:
- Merged news and Bitcoin price data by datetime, creating a final database for further analysis.
- Python scripts for data collection and processing.
- Notebooks with detailed analysis and visualizations.
- Examples of sentiment analysis and summarization workflows.
This repository is a resource for researchers and developers interested in financial data analysis, natural language processing, and cryptocurrency trends.