PyMC 5.8.0 is here, packed with some fantastic updates and improvements.
- Causal inference: added the do operator for modeling interventions
The do operator is going to be quite a big deal in terms of easing the workflow. You can now define a model, perform pearl-style interventions to create a counterfactual model and then do anything you would normally do: prior simulation, sampling, posterior simulation.
Check out the corresponding blog post, Causal analysis with PyMC: Answering "What If?" with the new do operator - PyMC Labs
- Added ICAR distribution
- Added JAX implementation for MatrixIsPositiveDefinite Op
…are many more exciting updates; view the summary of changes here
New Example NB:
Don’t forget to like this post and share it with your networks.