SmartBuildSim

SmartBuildSim simulates smart-building telemetry so you can explore forecasting, anomaly detection, clustering, and reinforcement learning workflows with a single toolkit.

Quick links

• Quickstart guide — step-by-step CLI walkthrough using the office-small scenario.
• examples/scripts/run_example.py — Python script orchestrating the entire workflow programmatically.
• examples/notebooks/smartbuildsim_workflow.ipynb — Jupyter notebook variant of the scripted workflow for interactive exploration.
• examples/configs/default.yaml — configuration used by the CLI examples and reference snippets.

Module reference

Dive into the dedicated reference pages for in-depth documentation and runnable examples:

• BIM schema
• Data generation
• Feature engineering
• Pipeline architektury
• Models: Overview, Forecasting, Anomaly detection, Clustering, Reinforcement learning
• Scenarios
• Command line interface
• Utilities
• Logging
• Visualisation

Typical experiment flow

1. Export or customise a BIM schema using smartbuildsim bim init.
2. Generate deterministic telemetry with smartbuildsim data generate.
3. Engineer features and train models for forecasting, anomaly detection, clustering, and reinforcement learning (models reference).
4. Visualise outcomes with smartbuildsim viz plot to verify anomalies and cluster assignments.
5. Iterate by applying configuration overrides (--override key=value) as shown in the CLI reference.

The quickstart expands each step and mirrors examples/scripts/run_example.py to ensure the documentation remains runnable end-to-end.

Build and preview the documentation

Install the documentation dependencies and serve the site locally before publishing to GitHub Pages:

pip install -r docs/requirements.txt
mkdocs serve

Build the static site (the output can be deployed to GitHub Pages):

mkdocs build --strict

The MkDocs configuration (mkdocs.yml) targets the docs/ directory and is ready for GitHub Pages once you push the generated site/ directory (or enable Pages on the repository).

This site is open source. Improve this page.