Installation

Recommended installation

For daily use:

pip install riskbands

If you also want the visual extras used in notebooks and Plotly demos:

pip install "riskbands[viz]"

To use the optional PySpark fit/transform paths:

pip install "riskbands[spark]"

The base package does not install PySpark. The spark extra uses pyspark>=3.5,<4.

Development environment

To work on the local repository, run tests, and execute notebooks:

git clone https://github.com/joaaomaia/RiskBands.git
cd RiskBands
pip install -e .[dev]

The dev extra adds useful tools for:

• tests with pytest
• notebooks with ipykernel
• .xlsx export
• Plotly visualizations
• release build and validation

Main library dependencies

The main installation covers the project core:

• pandas
• numpy
• scikit-learn
• optbinning
• optuna
• category_encoders

Practical notes

• .xlsx export requires a compatible engine such as openpyxl.
• The numeric supervised flow reuses optbinning.
• Optuna usage is optional.

Local documentation

The documentation site lives in docs-site/ and uses Astro + Starlight.

To run it locally:

cd docs-site
npm ci
npm run dev

To generate a production build:

cd docs-site
npm ci
npm run build

After installation

The best next step is usually:

1. Quickstart
2. Score and strategies
3. Outputs and diagnostics