By: Amy Qiu ([email protected]), Anthony Zhang ([email protected])
Inspired by the fintech challenge. We wanted to explore possible ways large-scale social trends could influence the market.
Sentigrade is constantly listening to tweets and analyzing the sentiments of messages about different companies. Over time, it builds up an idea of which companies are viewed positively, and which negatively.
Sentigrade also shows historical stock data, allowing users to look for potential relations.
A service constantly updates the database with information from the real-time Twitter message stream. It performs sentiment analysis and aggregates the result over fixed intervals.
The web backend, written in Flask, pulls Twitter data from the database and stock data from Yahoo Finance. The frontend is a simple jQuery/Bootstrap page to display the results in an informative, nice-looking way.
We originally intended to use arbitrary lists of items, retrieving every message from Twitter. However, this functionality was not available. Also, the stock data retrieval proved messy, though it worked well in the end.
Finishing the project ahead of schedule and getting to really flesh out the details.
Web development is scary to start with since you don't know where to begin, but once you hash out all the details, everything comes through.
Sentiment history. Actionable insights, possibly. Per-user settings for filtering, etc.