Mimir is a production-grade linear regression demand forecaster designed for the Parqcast ecosystem. It bridges the gap between historical data exports and automated, just-in-time inventory replenishment in Odoo.

Mimir reads Parquet exports generated by the Parqcast Odoo addon, runs a demand forecasting pipeline, and computes min/max replenishment rules. These optimized rules are then pushed back to Odoo stock.warehouse.orderpoint records to automate inventory management.

Currently supported Odoo versions: 18, 19

• Automated Demand Forecasting: Uses linear regression to project future demand based on historical data.
• Replenishment Rules Generation: Automatically computes minimum and maximum stock rules.
• Seamless Odoo Integration: Optionally pushes computed rules directly to stock.warehouse.orderpoint records via XML-RPC.
• Multilingual Support: Ships with UI translations for 19 languages out of the box.
• CLI Interface: Provides a lightweight CLI for headless execution and scripting.

Prerequisites: Python 3.12.3 (Required for Odoo.sh compatibility).

The project uses uv for dependency management.

# Install workspace + development dependencies
uv sync

# Run the test suite
uv run pytest tests/ -x -q

# Lint the codebase
uv run ruff check .

The addon (located in the odoo_addon/ directory) integrates directly into Odoo:

1. It reads ir.attachment exports generated by Parqcast.
2. It runs the forecasting pipeline.
3. It stores the forecast results and rules as attachments on a mimir.run record.
4. Optionally, it pushes these rules directly to stock.warehouse.orderpoint via XML-RPC.

For standalone execution or pipeline integration, use the CLI:

uv run mimir /path/to/parqcast/exports /path/to/output --verbose

A deterministic set of sample Parquet inputs can be generated for testing. The generated files are stored under test_data/ (which is git-ignored).

uv run python scripts/generate_test_data.py

The addon includes UI translations for 19 languages. All msgstr entries are fully populated (see odoo_addon/i18n/). Translation PRs are always welcome!

Built and maintained by DataStruct s.r.o. — an Odoo Official Partner based in the Czech Republic. We specialize in demand forecasting and ERP implementations for manufacturing, retail, and logistics in the Czech–German–Polish region.

• See it in production: Demand-forecast case study · Machine-builder materials: 67% → 94%
• Book a 30-min technical diagnostic: Schedule here
• Questions / Partnership: info@datastruct.tech

Companion Projects:

• parqcast (The primary exporter this tool reads from)
• parqcast-lambda
• parqcast-server

This project is licensed under the LGPL-3.0-or-later license.