Repository navigation
v2.0.0
v2.0.0 — Architectural Rewrite (Breaking Release)
⚠ Breaking Change: This is a complete architectural rewrite. Existing code using
v1.xwill need to be updated — see the migration guide below.
What's New
- New core engine built on Polars — significantly faster data processing for large journey datasets
- Revenue-based attribution — preprocessing, core, models, and reporting now support
conversion_valueas revenue instead of just conversion counts - New import path — the public API has been reorganised for clarity
- Updated and expanded test suite — unit tests rewritten to cover the new architecture
- Examples moved to
examples/— including usage of theconversion_value_as_frequencyparameter for Markov chains
Migration Guide
| v1.x | v2.0.0 | |
|---|---|---|
| Import | from marketing_attribution_models.MAM import MAM |
from mam.core import MAM |
| Python | >=3.8 | >=3.12 |
| Engine | pandas | Polars |
The legacy v1 code is preserved under codigo_legado/ for reference.
Pull Requests
- #103 — Nova MAM v2.0.0 — Architectural Migration
- #104 — Revenue-based attribution support
- #105 — CI: Trusted Publisher setup for PyPI
Full Changelog: v1.0.11...v2.0.0