We wanted to find a way to use open source machine learning resources to mine twitter for data about public figures. Polls can be biased because the people taking them are aware of what they are, and the people giving them may skew the results one way or another. Sentiment analysis may not be entirely accurate for individual tweets, but in aggregate, it gives a fairly clear picture of the sentiment of the country as a whole.

What it does

We built a heat map of the United States that displays positive and negatively ranked tweets as overlay points on a map. Each point makes the overlay more red or more green, depending on whether it's positive or not. This map can be filtered by keywords, some of which are helpfully provided in the left sidebar. There are quick-links there for congress members, and multi-searches for political parties (searches for 'republican', but also 'GOP', 'gop','(R)' etc.).

How we built it

We used python and the Flask framework for the back end of the website, and python (with the open source nltk module) to generate sentiment analysis for each tweet. The front end is css, html, javascript, and a few drop-in frameworks like Material Design Lite. We organized our source code with git and github, and our repository is availabale at: link

Challenges we ran into

We has a lot of difficulty configuring a webserver due to a few infuriatingly small errors early on, but we persisted, and the website is now available at It is also built automatically from a git repository, so no manual configuration is required to deploy an update- we just push to live and visit it in the browser.

Accomplishments that we're proud of

We're proud of the whole thing. We finished most of the basic features that we wanted ahead of schedule, and that left us enough time to add in vanity features like a logo and some cool statistics.

What we learned

Several of us had limited experience with python and git, but we learned how to use both (as much as you can in 24 hours), and figured out how to work together to create a solid project.

What's next for

The site is staying up for the foreseeable future while we discuss that.

Share this project: