Getting Started
Installation
You can install svy-sae from PyPI:
pip install svy-saeFor isolated environments, any standard workflow works (venv, uv, conda, etc.). If you already use uv, it integrates well with the project’s development setup.
Requirements
svy-sae requires Python 3.11 or newer.
Dependencies
svy-sae is built on a small set of core dependencies:
- JAX / jaxlib — accelerated numerical computing (CPU/GPU depending on your platform)
- SciPy — scientific routines used in optimization and numerics
- svy (
svy[report]) — survey design and direct estimation utilities used throughout workflows - tqdm — progress reporting for iterative procedures (e.g., bootstraps)
Usage
svy-sae provides implementations of both area-level and unit-level small area estimation workflows, along with supporting utilities for estimation, diagnostics, and uncertainty quantification.
Start with the Overview and Quick Tour sections to run an end-to-end example, then follow the model-specific tutorials for area-level and unit-level methods.
Development installation (optional)
If you are contributing or working from source, install the project in editable mode and include development dependencies:
pip install -e ".[dev]"Note: In this repository,
svymay be used as a local editable dependency during development.
License
See the LICENSE file in the repository for full terms.