DJ Panda

Quiz Interface
Post-Quiz Analysis
Song Recommendation

DJ Panda

An AI music companion for well-being and global awareness. Tell DJ Panda how you feel, and it uses real survey data to take you to a musical culture that fits, plays you a song, and explains exactly why.

Try it live · GitHub repo

Inspiration

Many people use music to manage their mood, but streaming recommendations are black boxes that rarely take listeners beyond familiar genres. For AI for Social Good, we wanted a recommender that is transparent about its picks, grounded in what real people report about music and their feelings, and that sparks curiosity about other cultures.

What it does

Three questions: how you feel (stressed, drained, lonely, restless, can't sleep, happy or energetic), how strongly, and what you want the music to do (calm down, lift your mood, get energized, or keep the vibe).
The DJ's read: DJ Panda shares what the survey says about people who feel like you and say music helps them.
Pick a world: a mood map and an interactive world map rank five music cultures by how well they fit: India, South Korea, Latin America, Europe and West Africa.
Now playing: a matching song with a Spotify preview, an animated DJ that dances to the beat, and a data-backed fun fact about the culture.
"Why this song?": a feature-by-feature comparison and a plain-language explanation of the pick.

How we built it

We combined three datasets: the Music & Mental Health survey, the Spotify Tracks Dataset, and the Culture-Aware Music Recommendation dataset (Last.fm, Hofstede dimensions and the World Happiness Report). Using Python and pandas, we cleaned the data, sorted songs into five cultural worlds, and built 84 mood profiles from survey respondents who said music helps them. Songs are matched to each profile by comparing standardized audio features like tempo, energy and valence. The front end is built with React, TypeScript, Vite, Motion and custom SVGs, and deployed with GitHub Pages. We also used Claude Code to help with coding and debugging.

Challenges we ran into

We struggled with choosing tools and coming up with a unique idea. As first-year students with limited coding experience, we chose the beginner-friendly Pop Culture dataset but initially struggled to see how it could solve a real-world issue.

Accomplishments that we're proud of

We are proud of how our team worked together to learn new software. Coming into our first datathon, we were unsure what to expect, but learning React helped us turn our idea into a working project.

What we learned

We learned how to uncover connections between datasets that may not have obvious links. Most importantly, we discovered how music can influence emotions and create meaningful moments of relief.

What's next for DJ Panda

Our next step would be publicly deploying the project. We could also expand DJ Panda by adding more songs from different regions and using more emotion-based data to help it better understand users.