Markada is a financial news sentiment analysis application developed to predict short term volatility of financial assets.
The goal of this app is to serve as a additional feature(check) to make a traiding decision either buy or sell a stock. This app functions as an API which can analyze an article given only the URL. The app will parse the article and return a sentiment score of either 0 with is negative or 1 which is positive.
You can find the app here: https://markada.herokuapp.com/ Sometimes heroku takes a little second to fire up so please allow a couple of seconds when first loading the app
You can find the analysis part of the app in this repo: https://github.com/casanas10/NewsSentimentAnalysis
- Python
- Flask
- Jinja2
- Python
- Natural language processing
- NLTK
- Pandas
- Numpy
- sklearn
- Classification
- beautifulsoup
- learned how to scrape news data from different news sources
- learned how to label news based on lexicon of negative/positive words
- learned how to clean text data and vectorize it for machine learning models
- trained and evaluated different machine learning models
- deployed the app to a cloud server as an API