faged-pipeline 0.1.3

FaGED pipeline: construct graph representation for open-space public buildings, extract prototypes, and compute (n)GED/FaGED for layout-behavior retrieval.

FaGED Pipeline

FaGED Pipeline is a research-oriented Python library for comparing spatial layouts and human behavioral patterns using function-aware graph representations.

It can also be used independently to construct graph representations of complex architectural layouts with open spaces and to extract layout prototypes from such configurations.

---

Key features

• Step-wise pipeline (Step0–Step5) for reproducible experiments
• Layout graph representation for complex open-space buildings
• Prototype extraction via community detection (e.g., Infomap)
• GED / nGED / FaGED computation + retrieval ranking outputs
• Designed for architectural research workflows

---

Installation

pip install faged-pipeline

---

Minimal example (layout graph → CaG)

from faged_pipeline import run_step2_then_step3, Step3Config

run_step2_then_step3(
    json_folder="DATA_ROOT/json",
    step2_output_folder="OUTPUT_ROOT/step2_basegraphs",
    step3_output_root="OUTPUT_ROOT/step3_transform/variants",
    step3_selected_folder="OUTPUT_ROOT/step3_transform/selected",
    step3_cfg=Step3Config(),
    save_png_step2=True,
    save_selection_csv_step3=True,
)

---

Full pipeline example (Step0–Step5)

The complete step-by-step notebook and documentation are hosted on GitHub:

• GitHub repository: https://github.com/songguoarchitect/faged
• Example notebook: quick_start.ipynb

Note: PyPI does not reliably render local images (relative paths).
We keep the PyPI page concise and place full visuals/examples on GitHub.

---

Data availability

A small example dataset (5 samples) is included for demonstration.
The complete dataset (150 images) is not publicly released due to data usage agreements.

---

Citation

If you use FaGED Pipeline in academic work, please cite:

@article{Guo2025FaGED,
  title   = {Function-Aware Graph-Based Layout Retrieval for Public Buildings},
  author  = {Guo, Song and Kee, Tris and Zhuang, Weimin},
  journal = {Automation in Construction},
  year    = {2025}
}

---

License

MIT License.