MoodSick
A group project for CIS 600 (Social Media and Data Mining) that explores how mood relates to music. It builds on the GTZAN music-genre dataset and a Spotify songs dataset, and runs as a set of containers on a shared Docker network.
architecture
Docker keeps everyone on the team on the same environment, so nobody hits dependency problems. The compose file creates a network called moodSickNetwork that connects four services.
| Service | Address |
|---|---|
| Frontend | localhost:3000 |
| Middleware | localhost:8080 |
| ML model | localhost:5000 |
| Database | localhost:27018 |
The middleware talks to the Spotify API, and the ML model reads its vectors from a hosted MongoDB Atlas database. Credentials live in per-service .env files, kept out of the repo.
run it
git clone https://github.com/ayushs-dev/MoodSick.git cd MoodSick/docker docker-compose build docker-compose up
data exploration
Before building the services, the team explored the data in Jupyter notebooks: an initial exploration of the datasets, a look at mood and genre dynamics on Spotify, and an autoencoder experiment.
Datasets: GTZAN (music genres) and a Spotify songs dataset, both from Kaggle.