The general idea of this is as follows:
Hackernews comments -> AI -> stock predictions
Don't expect it to give any useful predictions :).
It uses Steampipe to download Hackernews and Yahoo Finance data.
Only Hackernews comments are used. These comments are split up into 2+ word n-grams.
Linear regression is used to correlate the frequency of these n-grams over 1 day to percentage stock price changes on that day. A prediction is then made for the following day.
See conda.export for packages needed.
daily.py is the main script that is run daily.