Visualization in Bayesian workflow

By Jonah Gabry, Daniel P. Simpson, Aki Vehtari, Michael Betancourt, Andrew Gelman. Journal of the Royal Statistical Society: Series A (Statistics in Society), 2017

981 citations recorded by the source corpus.

Bayesian data analysis is about more than just computing a posterior distribution, and Bayesian visualization is about more than trace plots of Markov chains. Practical Bayesian data analysis, like all data analysis, is an iterative process of model building, inference, model checking and evaluation, and model expansion. Visualization is helpful in each of these stages of the Bayesian workflow and it is indispensable when drawing inferences from the types of modern, high dimensional models that are used by applied researchers.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Information Systems and Management

View Paper: https://www.semanticscholar.org/paper/fa9bc237dffe92361c42d21cf72592d479890bd4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Aki Vehtari
2. Andrew Gelman
3. Jonah Gabry
4. Yuling Yao
5. Osvaldo A. Martin
6. Colin Carroll
7. Ravi Kumar
8. Ari Hartikainen

In Computer Vision and Pattern Recognition

1. Jonah Gabry
2. Andrew Gelman
3. Aki Vehtari
4. Yuling Yao

In Information Systems and Management

1. Andrew Gelman
2. Aki Vehtari

Citations in the corpus

• Pareto Smoothed Importance Sampling (2015, background): “When working with many di ↵ erent models, the Stan team has noticed that high ˆ k values are also a useful indicator that there is something wrong with the data or the model (Gabry et al., 2019).”
• ArviZ a unified library for exploratory analysis of Bayesian models in Python (2019, methodology): “In addition to common plots for Bayesian analysis including a trace plot and forest plot, ArviZ implements other visualizations such as a plot for posterior predictive checks, a pair plot, and a parallel coordinate plot (Gabry et al. 2017).”
• Stacking for Non-mixing Bayesian Computations: The Curse and Blessing of Multimodal Posteriors (2020, methodology): “Stacking as part of Bayesian workflow We view stacking of parallel chains as sitting on the boundary between black-box inference and a larger Bayesian workflow (Gabry et al., 2019).”