seminrExtras 0.1.0

Python port of the seminrExtras R package: supplementary PLS-SEM evaluation and validation tools (CVPAT, PCM, cIPMA, COA, NCA, CTA-PLS, FIMIX-PLS, PLS-POS, congruence)

seminrExtras (Python)

Python port of the seminrExtras R package: supplementary tools for evaluating and validating PLS-SEM (partial least squares structural equation) models, built on the seminr estimation engine.

Every feature is validated for numerical parity against the R package on R-generated golden fixtures (1e-5 default tolerance; injected R random streams for resampling-based features).

Installation

pip install seminrExtras            # core
pip install "seminrExtras[plot]"    # + matplotlib for the plot layer
pip install "seminrExtras[pandas]"  # + pandas-friendly input

Requires Python >= 3.11 and seminr >= 0.2.1 (installed automatically).

Features

Mirroring the R package, evaluated over an already-estimated seminr model:

Feature	Functions
Cross-Validated Predictive Ability Test (CVPAT)	assess_cvpat, assess_cvpat_compare
Predictive Contribution of the Mediator (PCM)	assess_pcm
(Combined) Importance-Performance Map Analysis	assess_ipma, assess_cipma
Composite Overfit Analysis (COA)	assess_coa, predictive_deviance, deviance_tree, unstable_params, group_rules, competes
Necessary Condition Analysis (NCA)	assess_nca, assess_nca_esse
Confirmatory Tetrad Analysis (CTA-PLS)	assess_cta
FIMIX-PLS (latent-class segmentation)	assess_fimix, assess_fimix_compare
PLS-POS (prediction-oriented segmentation)	assess_pos, assess_pos_compare, pos_segments
Congruence testing	congruence_test

Usage

Specify and estimate a model with seminr, then evaluate it with seminrExtras:

from seminrExtras import assess_cvpat

result = assess_cvpat(model, nboot=200, seed=123)
print(result)              # readable table (R print parity)
result.summarize()         # detailed summary

# Plots (requires the `plot` extra):
from seminrExtras.plotting import plot
fig = plot(result)

Result objects are frozen dataclasses with a readable __str__, a summarize() method, and — via seminrExtras.plotting — matplotlib plots that mirror the R package's plot methods.

Development

uv sync                # install (dev group included)
uv run pytest          # tests
uv run ruff check .    # lint
uv run ruff format .   # format
uv run mypy src        # type-check (strict)

License

GPL-3.0-only, like the R package it ports.