nns-pm 0.1.0

Python port of NNS

pynns

Python port of the NNS (Nonlinear Nonparametric Statistics) R package.

Status

Alpha / 0.1.0 alpha. Parity-tested port of installed R NNS 12.0, with deferred paths documented below.

The package API is not stable yet. Current work is focused on matching the reference R implementation before adding higher-level Python conveniences. Native extension scaffolding has been removed; PyNNS is currently a pure-Python/NumPy/SciPy port.

Install

uv sync --group dev

Parity tests that miss the local cache require R plus the NNS R package:

install.packages("NNS")

Set PYNNS_OFFLINE=1 to force parity tests to use only the committed R cache.

Deferred paths:

• nns_var / nns_nowcast
• vectorized dy_d wrt for non-mean modes (obs, other eval points)
• dy_d eval_points="obs" parity gap (implemented but not yet R-parity compatible)
• direct raw-factor nns_m_reg
• boost threshold on stochastic path (n_features > 10)

The package is GPL-3.0-only and imports as pynns.

For user installs, the planned distribution package is nns-pm:

pip install nns-pm

Run Tests

uv run pytest
uv run ruff format .
uv run ruff check . --fix
uv run ruff check .
uv run mypy

Pytest uses 4 xdist workers by default. Override with PYNNS_PYTEST_WORKERS=<n> uv run pytest for larger or smaller machines.

Attribution

NNS was created by Fred Viole as the companion R package to Viole, F. and Nawrocki, D. (2013), Nonlinear Nonparametric Statistics: Using Partial Moments.

Upstream: https://github.com/OVVO-Financial/NNS

License

GPL-3.0-only