Soulmates!

Soulmates - Start NOW for FREE with local fellow Data Scientists!!
Soulmates - Get it while it's hot!

Inspiration

With the goal of crafting a personalized experience wrapped on top of meaningful analytics, Soulmates! is designed to bring people together. Datathons can be intimidating to first time hackers, so we sought to ease the transition for newcomers while providing a little something for the hardened veterans as well. Soulmates! encourage cross-cooperative behavior between attendees to help foster a productive learning environment.

What it does

Applicants can use our tool to explore personalized data tailored just for them. Given an userid, generated results include personalized welcome messages, details about how they fit in with TAMU Datathon, and graphs detailing other participants from the same university/college. Based on an individual's personal data, the algorithm matches them with the perfect 'Soulmate' from the applicant pool, as well as some potential besties. On top of this, possible team combinations are recommended to aid in the team building process.

How we built it

The team combination algorithm is built using K-means clustering while the 'Soulmate' radar recommends individuals with the highest degree of similarity to the user. Some additional data is added to the initial DataFrame, mainly the states where people are located to help match those of similar time zones, as well as the workshops people attended where incorporated as well.

Challenges we ran into

All three teammates are first time hackers, and two had almost no previous python experience. Sanitizing the initial data for comparisons took longer than expected, and time ran out before some goals could be reached. An alternative clustering algorithm was attempted that would prioritize those with fewer similarities with others to ensure they would be matched well but difficulties in implementation prevented the task from coming to fruition. This datathon was a learning process the whole way, and we came out stronger from it.

Accomplishments that we're proud of

We've learned a great deal participating in this hackathon, and we hope to take this knowledge foundation and build on it even further in preparation for the next one. We were able to submit a minimum viable product in time even if we weren't able to reach our stretch goals of testing more types of clustering algorithms and building a workshop recommendation system from the applicant data.

What we learned

For the development of this project, we learned about the pandas and matplotlib libraries, basic clustering algorithms, and the thrill of manipulating raw data.

What's next for Soulmates!

Besides implementing the stretch goals, the team wishes to learn some web-dev and deploy Soulmates! in the browser for a more friendly, ergonomic and personalized viewing experience to better emphasize the theme "for you" and better achieve its goal of brining people together.