brmspy 0.1.0

Pythonic interface to the brms R package using CmdStanPy

brmspy

Pythonic interface to R's brms for Bayesian regression modeling

brmspy brings the power of brms (Bayesian Regression Models using Stan) to Python, providing proper parameter names and seamless integration with the Python Bayesian ecosystem.

Quick Start

pip install brmspy

import brmspy

# One-time setup: install brms and CmdStan
brmspy.install_brms()

# Load example data
epilepsy = brmspy.get_brms_data("epilepsy")

# Fit model - returns arviz InferenceData by default
model = brmspy.fit(
    formula="count ~ zAge + zBase * Trt + (1|patient)",
    data=epilepsy,
    family="poisson",
    chains=4,
    iter=2000
)
idata = model.idata

# Analyze with arviz
import arviz as az
az.plot_posterior(idata)
az.summary(idata)

Key Features

• Proper Parameter Names: Returns b_Intercept, b_zAge, sd_patient__Intercept (not b_dim_0, sd_1_dim_0)
• Pythonic by Default: Returns arviz.InferenceData for seamless Python integration
• Flexible: Optional R brmsfit return for full brms functionality
• Formula Syntax: Use brms' intuitive formula interface
• Modern Stack: Python 3.8-3.14, brms + cmdstanr backend

Installation

# Install from PyPI
pip install brmspy

# Install with optional dependencies
pip install brmspy[viz]    # includes arviz, matplotlib
pip install brmspy[all]    # includes all optional dependencies

# First-time setup (installs brms and CmdStan in R)
python -c "import brmspy; brmspy.install_brms()"

Note: The package is imported as brmspy but installed as brmspy from PyPI.

Usage

Basic Model

import brmspy
import arviz as az

# Load data
kidney = brmspy.get_brms_data("kidney")

# Fit Gaussian model
idata = brmspy.fit(
    formula="time ~ age + disease",
    data=kidney,
    family="gaussian",
    return_type="idata"
)

# View summary
az.summary(idata)

With Priors

idata = brmspy.fit(
    formula="count ~ zAge + (1|patient)",
    data=epilepsy,
    family="poisson",
    priors=[
        ("normal(0, 0.5)", "b"),
        ("cauchy(0, 1)", "sd")
    ],
    return_type="idata"
)

Return Type Options

# Default: arviz InferenceData (Pythonic)
idata = brmspy.fit(..., return_type="idata")
az.plot_posterior(idata)

# R brmsfit object (for advanced R users)
fit = brmspy.fit(..., return_type="brmsfit")
import rpy2.robjects as ro
ro.globalenv['fit'] = fit
ro.r('summary(fit)')

# Both formats
result = brmspy.fit(..., return_type="both")
az.plot_posterior(result.idata)  # Python
# result.brmsfit for R methods

Sampling Parameters

model = brmspy.fit(
    formula="y ~ x",
    data=data,
    iter=2000,      # Total iterations per chain
    warmup=1000,    # Warmup iterations
    chains=4,       # Number of chains
    cores=4,        # Parallel cores
    seed=123        # Reproducibility
)

Requirements

Python: 3.8+

R Packages (auto-installed):

• brms ≥ 2.20.0
• cmdstanr
• posterior

Python Dependencies:

• rpy2 ≥ 3.5.0
• pandas ≥ 1.3.0
• numpy ≥ 1.20.0
• arviz (optional, for InferenceData conversion)

Development

# Clone repository
git clone https://github.com/kaitumisuuringute-keskus/brmspy.git
cd brmspy

# Setup environment (requires Python 3.8+)
./init-venv.sh

# Run tests
pytest tests/ -v

# Run tests with coverage
pytest tests/ --cov=brmspy --cov-report=html

License

Apache License 2.0

Credits

• Original concept: Adam Haber
• v0.1.0 modernization: Remi Sebastian Kits
• Powered by brms by Paul-Christian Bürkner