Statistical modeling in Python, with familiar R formulas

rparity fits mixed models, smooth effects and zero-inflated responses directly in Python. Start with a pandas data frame, write a formula, and carry the fitted model into prediction, hypothesis tests and estimated marginal means. Fitting and inference require no R installation or rpy2.

Install rparity | Run your first analysis | API reference | Source and examples

From your data to an answer

Suppose data contains repeated measurements with columns score, time and participant:

from rparity import Anova, emmeans, lmer, pairs

model = lmer("score ~ time + (1 | participant)", data)
print(model.summary())
print(Anova(model, type=2))

means = emmeans(model, "time", at={"time": [0, 5]})
print(pairs(means).summary())
predictions = model.predict(data, re_form="NA")

The quickstart includes a complete dataset and a runnable version of this workflow. Result tables are pandas data frames or named series; predictions can be used in your existing Python analysis.

Choose your analysis

You want to	Start here	Python API
Model repeated measurements or grouped observations	Mixed models	lmer, glmer
Estimate nonlinear effects without fixing their shape	Generalized additive models	gam, gam_check
Model overdispersed counts and excess zeros	Zero-inflated models	glmmTMB
Compare adjusted group means and treatment effects	Marginal means and contrasts	emmeans, pairs, contrast
Account for correlated or unequal residual variation	GLS and residual covariance	gls, corAR1, varIdent

Bringing an R analysis to Python?

The R-to-Python guide maps common lme4, mgcv, glmmTMB, emmeans, car and nlme workflows to rparity. Detailed migration pages compare formula syntax, method arguments and result extraction side by side.

Numerical agreement is checked against recorded R results. Read numerical accuracy and limitations for the supported scope and known differences. The Python implementation is MIT licensed.