Skip to content

Files

Latest commit

ffed3df · Jan 25, 2019

History

History
23 lines (18 loc) · 872 Bytes

README.md

File metadata and controls

23 lines (18 loc) · 872 Bytes

CoinMoodBot

Telegram bot which analyses text messages of crypto market participants in Twitter and recognizes their mood.
This analytics might be helpful to support investment decisions.

Pipeline

CoinMoodBot is deployed on Heroku server.
Main packages:

  • tweepy - to search for target messages containing key words
  • NLTK - to process text data
  • scikit-learn - to perform sentiment analysis on tweets
  • matplotlib - to plot sentiment analysis results

Other infrastructure tools:

  • PostgreSQL - to store sentiment analysis output and users' feedback
  • Amazon Web Services (AWS) - to store charts in PNG format
  • Chatbase - to collect data for analytics (users' logs)

Screenshots