-
-
Notifications
You must be signed in to change notification settings - Fork 2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Prediction on New Data Fails with Deterministic Variable #3346
Comments
You're right. It's a bit disappointing not to be able to store Currently, the only ways around this are:
|
This issue seems to be introduced in the recent With |
Tested this example with
|
… the name of a variable was in the supplied point, then the value for said variable would be taken from the point dictionary. This was too permissive, and different from what was done in _draw_value. There, the value could be taken from the point dictionary only if the variable had the attribute model, i.e. it was an RV instance instead of a Deterministic.
Thanks @dsvolk! I looked at the differences between the released |
* Fix for #3346. draw_values had a statement that said that if the name of a variable was in the supplied point, then the value for said variable would be taken from the point dictionary. This was too permissive, and different from what was done in _draw_value. There, the value could be taken from the point dictionary only if the variable had the attribute model, i.e. it was an RV instance instead of a Deterministic. * Changed expected SMC exact step values. * Added comment explaining why we check for the model attribute of parameters to get their values from the point dictionary.
I am not sure if this behavior is due to some theoretical limitation or not.....but the prediction of new data with sample_posterior_predictive() fails to recognize new new shared variables when the model was run with a Deterministic() parameter in the model. See example below from the quick start.
The text was updated successfully, but these errors were encountered: