This project is currently on hold due to the efforts required by my ongoing Ph.D.
Abu is an Ally for Bayesian StUdies.
Version : 0.0.5
Date : 06 Jul 2019
Mantainer: Felipe C. Argolo @argolof
Felipe Ferreira @fdemacedof
This software is alpha version. Use at your own risk.
Ab (أَب), from a theoretical, abstract form (آبَاءٌ ʼabawun) (triliteral ʼ-b-w) is Arabic for "father".
In the construct state, Abū (أبو) is followed by another word to form a complete name, e.g.: Abu Mazen, another name for Mahmoud Abbas.
Abu may be used as a kunya, an honorific. To refer to a man by his fatherhood (of male offspring) is polite, so that ʼabū takes the function of an honorific. 1
Abu aids with Bayesian Inference procedures.
For an intuitive introduction about it, I recommend this text.
Abu offers mathematical tools and visualization schemes that mirror abstractions used in Bayesian analysis. It exports Interact.jl widgets.
* Interactive simulations of restricted scenarios (e.g. conjugate distributions). It explores *posterior* sensitivity to user choices (e.g. hyperparameters and family of *prior distribution*)
* Gaussian conjugates with known precision.
* ( ) Diagnostics for Convergence criteria and Goodness of Fit (e.g. WAIC)
* ( ) Scaffolded pipelines of analysis for common models (e.g. Gaussian Processes)
Check the specific documentation for contributions and releases. We welcome developers with any level of background expertise.
You can read an introduction to the philosophy and basic applications of Bayesian inference here (undergoing translation). Check the book Statistical Rethinking: A Bayesian Course with Examples in R and Stan (Chapman & Hall/CRC Texts in Statistical Science) for in depth content.
The initial module will offer capabilties similar to shinystan along with a few more utilities.
