Our goal during this hackathon was to build a data-heavy hack. We realized, in fact, that devpost was an incredible source of data! With 56,000 total projects available to analyze, this was the perfect direction for us.
What it does
We decided to build a project that uses past hackathon project data to inform and help hackers. Specifically, our hack:
- Data visualizes past data.
- Generates project name / project idea given past projects, and
- Determines your project's chances of winning a prize.
How we built it
Our web application stack includes the following:
- API Layer: We wrote a Python Flask API layer to communicate with our front-end.
- Analytics: We leveraged Python for all of the data scraping and manipulation into proper formats for the front-end.
markovify: We used this library to, given a corpus of past projects and names, generate project names and projects.
- IBM Watson: We used this technology to tag projects with keywords. We also wrote our own, separate method for keyword generation. Using this, we're able to understand how a project ranks against others in the context of winning.
Challenges we ran into
Key challenges included:
- API Keys: we were limited by the number of API keys available, so we had to be creative in how to use them and how to substitute our own techniques.
- Key libraries: we realized that libraries such as
markovifywere layers of abstraction that, without a clear understanding of how they work, were not 100% useful. We had to make sure we understood how and why those libraries worked and how we could shape the input for the best output.
Accomplishments that we're proud of
We're really excited that we have a full web application that has a ton of potential. However, the best moment came when the following randomly-generated project was outputted, showing us that even one project created thanks to our generator would be worth it:
jane is a platform that serves up real-time information on bed availability and predictions on when their shelter will fill up in an open format that all can use, via public apis.
What we learned
We learned a ton of new technologies and were challenged with a really cool data-driven project. In the future, we're really excited to push our limits in this space.
What's next for PostDev
We'd love to ship this as a web app in the near future!