plotly web app

Interactive Dash app for exploring how different positive/negative score splits across multiple pods affect per-pod and average ROC-AUC.

Python and environment management

This project now targets Python 3.12 and uses uv as the default workflow.

Initialize the environment

uv python install 3.12
uv sync

uv sync creates a local .venv/ and installs the package plus the minimal direct runtime dependencies declared in pyproject.toml.

Direct runtime dependencies

The project keeps only the libraries it imports directly at runtime:

• dash
• gunicorn
• numpy
• plotly
• scikit-learn

Transitive packages such as Flask, SciPy, and Werkzeug are installed automatically through those top-level dependencies.

Run locally

Live version

This version recomputes the split distributions and ROC-AUC values on each slider update.

uv run python app.py

Precomputed version

Generate the cached figures and ROC-AUC combinations first:

uv run python create_content.py
uv run python app_2.py

app_2.py is the faster variant and the default deployment target.

Deployment

Gunicorn

uv run gunicorn app_2:server -b 0.0.0.0:8050

Google App Engine

app.yaml is configured for Python 3.12.

gcloud init
gcloud app deploy ./app.yaml -Y

Docker

Build the image:

docker build -t plotly_app:1.0 . --build-arg runtime_path="."

Run it locally:

docker run --rm --name dash_app -d -p 80:80 plotly_app:1.0

If you want to run the published image on a VM:

chmod +x vm_docker_setup.sh
./vm_docker_setup.sh