Getting Started

AMMM is a Bayesian marketing mix modelling library built on PyMC, with a V2 driver architecture and a stage-gated workflow. Every run produces ordered artefacts (from metadata and pre-diagnostics through optimisation and interpretation), so you can audit model quality before acting on outputs.

Recommended path for new users:

1. Install and verify the environment.
2. Run the quickstart pipeline end to end.
3. Prepare your own data correctly.
4. Configure model and optimisation settings for your use case.

• Installation: Environment setup, editable install, and CLI/API verification.
• Quickstart: First full run with stage-based output inspection and diagnostics checks.
• Data Preparation: Input schema, validation expectations, and pre-diagnostics behaviour.
• Configuration Guide: Main YAML options, run modes, and optimisation settings.