@pymc-devs The insights you shared were really valuable and helped clarify a lot of concepts around MMM. I’ve been diving into MMM using the PyMC-Marketing MMM class, and I’m curious about setting customized priors for my model. Specifically, I want to know how I can set different priors for media channels (e.g., TV, digital ads) and control variables (e.g., seasonal effects, events like promotions). What’s the best way to approach this in terms of distribution choices and values for priors? Should I use a specific prior for each marketing channel and control variable, or is it better to keep them uniform? Any advice on how to tailor the priors to reflect the business context would be really helpful!
![Seungmin "그렇게, 천천히, 우리(As we are)" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/kAzmhLHePqU/mqdefault.jpg)
Amazing Presentation and Explanation. Congratulations
@pymc-devs The insights you shared were really valuable and helped clarify a lot of concepts around MMM. I’ve been diving into MMM using the PyMC-Marketing MMM class, and I’m curious about setting customized priors for my model. Specifically, I want to know how I can set different priors for media channels (e.g., TV, digital ads) and control variables (e.g., seasonal effects, events like promotions).
What’s the best way to approach this in terms of distribution choices and values for priors? Should I use a specific prior for each marketing channel and control variable, or is it better to keep them uniform? Any advice on how to tailor the priors to reflect the business context would be really helpful!
truly useful and thanks for sharing this! is there any code or use case?
an insightful presentation, kudos to you guys!👏👏 Any suggestions on how to built this, a media mix model from scratch using lightweightMMM in python?
Link to Jin and colleagues, please.
Here is a link to the original paper, Bayesian Methods for Media Mix Modeling with Carryover and Shape Effects , research.google/pubs/pub46001/
@@pymc-devs Thank you.
Can you share that code at min. 19 in a GitHub repository? Thanks!