Open aleicazatti opened 4 months ago
We should also create specific issues for each distribution, that will help us distribute the work and we could use the specific issues to collect and discuss the specific information we want to add to each element in the grid.
For the See Also, we could use this as a guide https://upload.wikimedia.org/wikipedia/commons/a/af/ProbOnto2.5.jpg, after a quick look I think a good principle could be to add a "See also" if two distributions are first neighbours in this graph. We could consider other cases too, for instance it makes sense that NegativeBinomial and ZINB should have mutual "see also" links. And also we should consider simplifications, for instance merge Normal with StandardNormal, or StudentT 1, 2 and 3 together with NoncentralT1. Still it could be a good guide.
Other sources for stories or useful information can be https://distribution-explorer.github.io/discrete/bernoulli.html and https://www.acsu.buffalo.edu/~adamcunn/probability/betabinomial.html
Progress tracker for individual distribution pages:
I'm working on the Student's T
I will continue with the logistic distribution
Working on Cauchy
Working on lognormal
Working on exponential
Working on Poisson
Working on continuous uniform
working on AsymetricLaplace
Working on ChiSquared
working on Gamma
working on ExGaussian
Working on HalfCauchy
Working on HalfNormal
Working on HalfStudentT
Working on InverseGamma
Working on Kumaraswamy
Working on Laplace
Working on Log-Logistic
Working on LogitNormal
Tell us about it
The current presentation of distributions on the PreliZ API page is too verbose, which is suitable for technical details but not very user-friendly.
Thoughts on implementation
1. Create a distributions grid page:
Develop a new page with a grid of images (like ArviZ example gallery), each representing a distribution's probability density function (PDF).
Clicking on an image should lead to more detailed information, including the "story" or history of the distribution.
2. Detailed individual pages for each distribution:
Simplify the information by distilling content from sources like Wikipedia, making it more accessible.
Include concise, "Bayesian-focused" stories for each distribution.
Provide visual examples, parameter explanations, and common uses.
Implement "See also" sections to link related distributions.
Add tags for easier searching based on domain, parameters, symmetry, etc.