Adding a penalty factor during initial model build will not allow for a model refresh

[jsabastian]

Project Robyn

Describe issue

Setting the experimental parameter 'add_penalty_factor = TRUE' during the initial model build will not allow the 'check_hyper_fixed()' function to pass, resulting in the following error:

Error in check_hyper_fixed(InputCollect, dt_hyper_fixed, add_penalty_factor) :
Input 'dt_hyper_fixed' is invalid. Please provide 'OutputCollect$resultHypParam' result from previous runs or 'pareto_hyperparameters.csv' data with desired model ID

Robyn Open Source MMM Users Post

Link

Environment & Robyn version

Make sure you're using the latest Robyn version before you post an issue.

• Robyn 3.8.1
• R version 4.2.2

[laresbernardo]

Correction on description to be able to replicate this bug: initial model runs OK, but when refreshing with robyn_refresh() user gets posted error.

[laresbernardo]

Ok, I've just pushed a fix. Given it's an experimental feature, it's expected to suffer some changes such as these: now the names of the hyperparameters for penalization ends (not starts anymore) with "_penalty" to keep the same format as the other hyperparameters (and show up in the prints nicely like attached image).

If you can please update to latest dev version, recreate your original model, export JSON file and run robyn_refresh() it should run smoothly. Please, let me know if that is or isn't the case so we can FUP or close this bug. Thanks @jsabastian

[jsabastian]

@laresbernardo Rebuild of the model using the penalty parameter refreshed correctly. Thanks for the fix! Also, I like the pareto front progress bar!

[laresbernardo]

Thanks, glad you found it useful. And thanks for checking back, I'll close this thread but feel free to reopen it if you get any further issues or comments.