Judgd - Deep Learning from Devpost
Aaron and I were wondering what it definitively takes to win a hackathon. As in, is there some means by which you can have a guaranteed at any hackathon if you just pick the right set of technologies and execute on it well?
We found the answer, and it's yes.
So what does this project do? This project starts by scraping every single hackathon project that's ever been put into Devpost and then attempts to derive some value from that information by putting them into a Deep Neural Network (DNN). At the time of this writing the data set that we are working with is ~52,000 projects and ~72,000 Github contributors to these hackathon projects.
We begin by grabbing all of this data, then building a best-fit linear model to it, and then allow you to make some predictions upon which you can test their viability.
You can read about the rest on Github.
Things we Learned
TensorFlow is really hard.