Module for Custom Priors and Distributions #322
Replies: 1 comment
-
Hi Steven, loc_range = range(250, 1000, 200) Loop through each combination of loc and scalefor loc in loc_range:
Please suggest if you have any insights regarding this. |
Beta Was this translation helpful? Give feedback.
-
Hello everyone,
I've been working on setting up custom priors for the lightweight_mmm model, specifically for the hill-adstock model. After some effort, I have finally figured out a way to re-assign the distributions and parameters effectively. I'm planning to create a Pull Request (PR) to contribute this to the open-source effort, but I wanted to share the implementation here for immediate use for those looking for a solution.
Custom Priors Implementation
Here is how you can define custom priors using different distributions:
From here, you can run the fit() method on custom_mmm as you would with the normal model, and it will use the custom priors defined in the dictionary above. Please let me know if there's interest in having this as part of the package itself for people to use consistently. I'll be happy to proceed with creating a PR for this contribution.
Beta Was this translation helpful? Give feedback.
All reactions