Bibat: batteries-included Bayesian analysis template
Keywords: Bayesian statistics, workflow, automation, template, data science tooling
TL;DR: A template and Python library that you can use to write a scalable Bayesian statistical analysis project
Abstract: Bayesian statistical workflow offers a powerful way to learn from data,
but software software projects that implement complex Bayesian workflows
in practice are unusual, partly due to the difficulty of orchestrating
Bayesian statistical software. Bibat addresses this challenge by providing
a full-featured, scalable Bayesian statistical analysis project using
an interactive template. Bibat is available on the Python Package index,
documented at <https://bibat.readthedocs.io/> and developed at <https://github.com/teddygroves/bibat/>. Bibat is free to use under the MIT license. This paper explains the motivation for bibat, describes intended usage, discusses bibat's design, compares bibat with similar software, highlights several examples of bibat's use in science and provides links to community resources associated with bibat.
Submission Checklist: Yes
Broader Impact Statement: Yes
Paper Availability And License: Yes
Code Of Conduct: Yes
Optional Meta-Data For Green-AutoML: This blue field is just for structuring purposes and cannot be filled.
Steps For Environmental Footprint Reduction During Development: avoid unnecessary computation with makefile specification, use efficient storage formats, avoid running excessively long MCMC, run computations locally on commercial cpus
CPU Hours: 0.5
GPU Hours: 0
TPU Hours: 0
Estimated CO2e Footprint: 0.02
Community Implementations: https://github.com/teddygroves/bibat
Submission Number: 5
Loading