pymc-devs/pymc

Add core notebook on sample_posterior_predictive for out-of-model predictions

Open

#6812 opened on Jul 4, 2023

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Python (7,926 stars) (1,902 forks)batch import
docshelp wanted

Description

Description

We wrote a series of examples for the PyMC labs blogpost: https://www.pymc-labs.io/blog-posts/out-of-model-predictions-with-pymc/

It would be good to write a more technical companion that explains exactly how sample_posterior_predictive works and why it works this way, and when is it valid/not valid to use if for out-of-model predictions.

There is a long thread here that highlights aspects of the current behavior that are unclear / surprising for most users (including developers): https://github.com/pymc-devs/pymc/issues/6047

Note: This is not a beginner friendly issue.

Contributor guide

Tech stack
python
Domain
documentationdata
Issue type
documentation
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
3-5 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Familiarity with PyMCUnderstanding of posterior predictive samplingRead issue #6047
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
10
Research direction
Review the blog post at the provided link and the discussion in issue #6047 to understand common confusions. The notebook should explain how sample posterior predictive works internally, when it is valid for out of model predictions, and include examples and caveats. No existing work has been started.