API Overview

This page summarises the AMMM V2 API surface and the recommended entrypoints.

Public Runtime Entry Points

CLI

Primary script:

python runme.py

See Configuration Reference for full CLI flags.

Driver API

Preferred import:

from driver import MMMBaseDriverV2

Compatibility import:

from src.driver import MMMBaseDriverV2

Minimal usage:

from driver import MMMBaseDriverV2

driver = MMMBaseDriverV2(
    config_filename="data-config/demo_config.yml",
    input_filename="data-config/demo_data.csv",
    holidays_filename="data-config/holidays.csv",
    results_filename="results",
)
driver.main()

Package-Level Modules

• Core: model classes, transformations, optimisation primitives.
• Driver: workflow orchestration and run-time operations.
• Preprocessing: config parsing, data preparation, validation.
• Sketch: plotting and reporting helpers.
• Top-Level Utilities: shared utility functions.

Workflow and Artefact Model

AMMM V2 writes outputs to stage folders:

• 00_run_metadata/
• 10_pre_diagnostics/
• 20_model_fit/
• 30_model_assessment/
• 40_decomposition/
• 50_diagnostics/
• 60_response_curves/
• 70_optimisation/
• 80_interpretation/

See:

• Output Files
• Output Schema

Diagnostics Gate Fields

Machine-readable fields for downstream checks:

• 50_diagnostics/convergence_report.json -> converged
• 50_diagnostics/calibration_report.json -> well_calibrated
• 50_diagnostics/pareto_k_summary.json -> ok

diagnostics_gating: strict|warn|off controls pipeline behaviour on failed convergence diagnostics (strict halts).

Caveat

Diagnostics and calibration checks assess computational and predictive adequacy; they do not, on their own, establish causal validity.