Welcome to Visual Inference Software Development project!

Convenient and readily available software is lacking for visualizing the process of statistical inference for various applications in statistical practice. The "logic" of inference, should be clearly depicted, if possible.

The project summary page you can find here.

The project is to enable and encourage visual depictions of inference at work - the process of learning from observations by their explicit representation as outcomes generated from unknown (in some aspects) probability models.

In the Bayesian paradigm, this is very directly seen in the posterior=prior*likelihood in the full dimension of the parameters, with direct averaging to any sub-dimension of the parameter space.

In the frequency paradigm this can start with likelihood=posterior/prior in the full dimension (where it is invariant to the prior) but there are perhaps unresolvable challenges in focussing down to any sub-dimension.

ToyExamples and _failed_ journal submission available.