shinyStan: a graphical user interface for exploring Bayesian models after MCMC

We have an excellent talk that combines two great aspects of R: Shiny and Stan.

We would like to thank The Orchard (http://www.theorchard.com/splash/) for hosting us this month. Please note the new location is at East 4th Street between Lafayette and Bowery.

About the Talk:

We will present shinyStan (https://github.com/jgabry/SHINYstan) an R package and Shiny (http://shiny.rstudio.com/) app providing a graphical user interface for exploring Bayesian models estimated by Markov chain Monte Carlo (MCMC) methods. In addition to demonstrating shinyStan we'll also discuss the importance of making tools for model exploration (visualizing, checking and diagnosing, comparing, inferring and predicting) easier to use and more integrated with software for model building and computation.

Most applied Bayesian data analysis requires employing a Markov chain Monte Carlo (MCMC) algorithm to obtain samples from the posterior distributions of the quantities of interest. Diagnosing convergence, checking the fit of the model, and producing graphical and numerical summaries of the parameters of interest is an important but often laborious process that slows the down the creative and exploratory process of model building.

shinyStan, a package for R and Shiny app, is designed to facilitate this process in two primary ways:

1. Providing interactive visual model exploration:

shinyStan provides immediate, informative, customizable visual and numerical summaries of model parameters and convergence diagnostics for MCMC simulations. Although SHINYstan was initially envisioned to be a companion to the RStan (http://mc-stan.org/rstan.html) package (the R interface to the Stan programming language for Bayesian statistical inference), it can also be used to explore the output from any other program (e.g. Jags, Bugs, SAS) or any user-written MCMC algorithm.

2. Making saving and sharing more convenient:

shinyStan allows you to store the basic components of an entire project (code, posterior samples, graphs, tables, notes) in a single object. Users can also export graphics into their R sessions as ggplot2 objects for further customization and easy integration in reports or post-processing for publication.

About Jonah:

Jonah Gabry is a graduate student at Columbia University (http://stat.columbia.edu/) studying statistics and concentrating on applications in the social sciences. He is currently researching statistical models for historical social scientific inquiry. Before studying statistics he used to study Spanish literature and was a touring musician (http://www.allmusic.com/artist/jonah-sol-gabry-mn0002601363/credits) for a (very) hot second.

About The Orchard:

The Orchard is a pioneering music, video and film distribution company and top-ranked Multi Channel Network operating in more than 25 global markets. We have a robust technology practice that includes Product Managers, QA Engineers and Software Engineers who work diligently to provide our clients with cutting edge solutions. Some of our most exciting projects focus on data-driven insights and analytics tools as well as optimization of monetization and reach for our YouTube business.

Pizza will begin at 6:30, announcements and the talk start at 7 and we will head to the bar after.