A simple cryptonews sentiment classification app.
The model (simple tf-idf & logreg) is trained with ~4500 news titles manually classiified into positive, neutral or negative.
Launching the app:
- install requirements from
requirements.txt
- execute
streamlit run src/app.py
To run this as a background process, you can do (streamlit run src/app.py > streamlit.log 2>&1 &)
. This will also write all Stramlit logs including error logs to the file streamlit.log
.