dataviz-projects

Dance styles analytics dashboard

Multi-linked Plotly Dash dashboard for exploring a global dance-styles dataset: parallel coordinates, Sankey flow, geo bubbles, and time-period small multiples. Built as a small, installable Python package with health checks and environment-based configuration.

Live demo: ishelar-dance-analytics.hf.space · Space: huggingface.co/spaces/ishelar/dance-analytics

Problem

Researchers and enthusiasts comparing traditions need linked views (filter in one chart, see the effect elsewhere) without juggling separate tools. This app wires PCP brushing and Sankey clicks into consistent cohorts and map context.

Features

• Linked filtering: parallel-coordinates ranges and Sankey node clicks narrow Sankey + PCP
• Geo bubble map with country focus dropdown
• Small multiples by coarse time-period group
• Print-oriented layout (A4 landscape CSS in assets/custom.css)
• GET /health — liveness
• GET /ready — dataset and centroids files on disk

Tech stack

• Python 3.11+
• Dash, Plotly, Pandas, NumPy
• Gunicorn (production WSGI)
• python-dotenv (optional local .env)

Quick start (local)

From this directory (dance-analytics-dashboard/):

python3.11 -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -U pip
pip install -e ".[dev]"
cp .env.example .env          # optional; defaults match repo layout
pytest
dance-dashboard               # or: python -m dance_analytics.main

Open http://127.0.0.1:8051 (or the PORT you set).

Sanity checks without opening the browser

curl -s http://127.0.0.1:8051/health
curl -s http://127.0.0.1:8051/ready

Configuration

Variable	Purpose
DANCE_ANALYTICS_DATA_PATH	CSV path (absolute or relative to app root)
DANCE_ANALYTICS_CENTROIDS_PATH	JSON map of country → [lat, lon]
HOST / PORT / DEBUG	Dev server binding
LOG_LEVEL	Python logging level (INFO, DEBUG, …)
DANCE_ANALYTICS_PAGE_TITLE	Browser tab title

See .env.example.

Production-style run (Gunicorn)

After pip install -e . from this directory:

gunicorn "dance_analytics.main:server" -b 0.0.0.0:8051 --workers 2

Run with Docker

docker build -t dance-analytics .
docker run -p 7860:7860 dance-analytics
# open http://127.0.0.1:7860

The same Dockerfile is what powers the live Hugging Face Space — non-root uid 1000, port 7860, gunicorn (2 workers), HEALTHCHECK on /health.

Project layout

dance-analytics-dashboard/
  Dockerfile             # Hugging Face / generic container build
  .dockerignore
  pyproject.toml
  README.md
  .env.example
  data/
    original_dance_dataset.csv
    country_centroids.json
  assets/
    custom.css
  src/dance_analytics/
    main.py              # app factory + /health + /ready
    config.py
    data/                # load + schema + transforms
    charts/              # figure builders
    dashboard/           # layout + callbacks + shared layout constants
  tests/

Example usage

1. Start the app (dance-dashboard).
2. Drag ranges on the parallel coordinates plot — Sankey and PCP update to the filtered subset.
3. Click a Sankey node — same behaviour.
4. Pick a country in the map card dropdown — map zooms to stored centroids.

Roadmap

• Row-level schema validation (e.g. pydantic or pandera)
• Precomputed aggregates for very large CSVs
• Centroids from a maintained geo dataset instead of static JSON
• CI: build the Docker image on every push and smoke-test /ready before deploy

License

MIT — see LICENSE.

---

Resume one-liner: Python/Dash analytics dashboard with linked cross-filtering, geospatial aggregation, health/readiness endpoints, and env-driven configuration suitable for portfolio demos.

This site is open source. Improve this page.