Sentiment Analysis on News to Predict Stock Performace

How to run:

1. Clone the project from git
2. Navigate to 'outer_folder' directory
3. docker-compose up
4. In a new terminal enter into the docker shell, docker exec -it <container_name> sh
5. Start up the celery beat scheduler: celery -A mysite worker --beat -S django -l info

Tasks

Finished

• news scraper (get_stock_articles.py)
• content scraper (get_content_from_articles.py)
• django models created
• stock historical data theough Yahoo Finance API (get_historical_stock_data.py)
• initial views and gainer graphs
• dockerfile and docker-compose

ToDo

• checkbox to select stocks to study
• auto-add of stocks to study (from gainers/ losers)
• host site
• automated task schedule for fetching & storing articles and price data
• possible cleanup tasks of data after certain period of time or after sentiment calculation etc.
• sentiment analysis (sklearn_sa.py)
• ground truth for articles summary (positive or negative, binary)
• method for comparing senitment analysis results with ground truth
• alert mechanisms for gaining stocks (auto-email)