The value of the budget was miscalculated

[Darren-Li]

https://www.pymc-marketing.io/en/stable/notebooks/mmm/mmm_case_study.html#budget-optimization

the budget should be the overall spend for all channels in one week. so the code should be
budget = X_test[channel_columns].mean(axis=0).sum()
but not the everage spend for one channel

also, the code for the chat in the image is wrong as well, And, obviously, it doesn't match the picture below.

The same goes for the next one.

mean_spend_per_period_test = (
    # X_test[channel_columns].sum(axis=0).div(num_periods * len(channel_columns))
    X_test[channel_columns].mean(axis=0)
)

[williambdean]

Hi, Thanks for bringing to our attention!
Is this also present in the latest docs? https://www.pymc-marketing.io/en/latest/notebooks/mmm/mmm_case_study.html#budget-optimization

[Darren-Li]

Thank you for your prompt reply:)

From your shared page(dev verison), I also have some points, you can double check again.

1. For code snippet below. all_budget_per_channel.div(num_periods * len(channel_columns)) should be changed to all_budget_per_channel.div(num_periods )
   

2. The same for those snippets below. It should not be divided by len(channel columns)
   
   

3. I got this error when run the below code AttributeError: 'MMM' object has no attribute 'optimize_budget', maybe it is new function in an upcoming update. But one of my doubts is budget=per_channel_weekly_budget,, is it right, or it should be all_budget #Total budget for the test set ?
   

[Darren-Li]

Can I find out when a version update for MMM is coming?

[williambdean]

We are working on an official release. However, you can pip install the main branch from github

[williambdean]

CC @cetagostini

[Darren-Li]

We are working on an official release. However, you can pip install the main branch from github

I will wait for the release, I am afraid of making a mistake, ha ha ha, I am not very familiar with these operations.