Stan is a probabilistic programming language for statistical inference written in C++. The Stan language is used to specify a (Bayesian) statistical model with an imperative program calculating the log probability density function. Stan is licensed under the New BSD License. Stan is named in honour of Stanislaw Ulam, pioneer of the Monte Carlo method. Stan was created by a development team consisting of 52 members that includes Andrew Gelman, Bob Carpenter, Daniel Lee, Ben Goodrich, and others.
Example A simple linear regression model can be described as y n = α + β x n + ϵ n {\displaystyle y_{n}=\alpha +\beta x_{n}+\epsilon _{n}} , where ϵ n ∼ normal ( 0 , σ ) {\displaystyle \epsilon _{n}\sim {\text{normal}}(0,\sigma )} . This can also be expressed as y n ∼ normal ( α + β X n , σ ) {\displaystyle y_{n}\sim {\text{normal}}(\alpha +\beta X_{n},\sigma )} . The latter form can be written in Stan as the following:
Interfaces The Stan language itself can be accessed through several interfaces:
CmdStan – a command-line executable for the shell, CmdStanR and rstan – R software libraries, CmdStanPy and PyStan – libraries for the Python programming language, CmdStan.rb - library for the Ruby programming language, MatlabStan – integration with the MATLAB numerical computing environment, Stan.jl – integration with the Julia programming language, StataStan – integration with Stata. Stan Playground - online at [1] In addition, higher-level interfaces are provided with packages using Stan as backend, primarily in the R language:
rstanarm provides a drop-in replacement for frequentist models provided by base R and lme4 using the R formula syntax; brms provides a wide array of linear and nonlinear models using the R formula syntax; prophet provides automated procedures for time series forecasting.
Algorithms Stan implements gradient-based Markov chain Monte Carlo (MCMC) algorithms for Bayesian inference, stochastic, gradient-based variational Bayesian methods for approximate Bayesian inference, and gradient-based optimization for penalized maximum likelihood estimation.
MCMC algorithms: Hamiltonian Monte Carlo (HMC) No-U-Turn sampler (NUTS), a variant of HMC and Stan's default MCMC engine Variational inference algorithms: Automatic Differentiation Variational Inference Pathfinder: Parallel quasi-Newton variational inference Optimization algorithms: Limited-memory BFGS (L-BFGS) (Stan's default optimization algorithm) Broyden–Fletcher–Goldfarb–Shanno algorithm (BFGS) Laplace's approximation for classical standard error estimates and approximate Bayesian posteriors
Automatic differentiation Stan implements reverse-mode automatic differentiation to calculate gradients of the model, which is required by HMC, NUTS, L-BFGS, BFGS, and variational inference. The automatic differentiation within Stan can be used outside of the probabilistic programming language.
Usage Stan is used in fields including social science, pharmaceutical statistics, market research, and medical imaging.
See also PyMC is a probabilistic programming language in Python ArviZ a Python library for Exploratory Analysis of Bayesian Models
References
Further reading Carpenter, Bob; Gelman, Andrew; Hoffman, Matthew; Lee, Daniel; Goodrich, Ben; Betancourt, Michael; Brubaker, Marcus; Guo, Jiqiang; Li, Peter; Riddell, Allen (2017). "Stan: A Probabilistic Programming Language". Journal of Statistical Software. 76 (1): 1–32. doi:10.18637/jss.v076.i01. ISSN 1548-7660. PMC 9788645. PMID 36568334. Gelman, Andrew, Daniel Lee, and Jiqiang Guo (2015). Stan: A probabilistic programming language for Bayesian inference and optimization, Journal of Educational and Behavioral Statistics. Hoffman, Matthew D., Bob Carpenter, and Andrew Gelman (2012). Stan, scalable software for Bayesian modeling Archived 2015-01-21 at the Wayback Machine, Proceedings of the NIPS Workshop on Probabilistic Programming.
External links Stan web site Stan source, a Git repository hosted on GitHub
