Skip to content
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

Linear vs multiplicative #96

Closed
fedemendez1 opened this issue Jun 21, 2021 · 1 comment
Closed

Linear vs multiplicative #96

fedemendez1 opened this issue Jun 21, 2021 · 1 comment

Comments

@fedemendez1
Copy link

Hi team,

What do you think about trying a multiplicative model instead of linear to capture synergy between channels? Is this something you have considered? Interested to hear your thoughts on that.

family=gaussian(link="log")

instead of

family = "gaussian",

Thanks,
R

@Leonelsentana
Copy link
Contributor

Hey Retiuk,

Thanks for your interest in Robyn :) We have considered multiplicative models in the past. However, we aimed for easy-to-interpret and computing-efficient models. This is because democratizing, scaling and inspiring marketing models is our main goal. If you have a lot of variables and sub-variables (50 different FB, 50 different TV variables, 50 different affiliate variables, etc.), you get high-cardinality, a lot of multiplication and therefore computationally unreasonable models, which may collide with an easy-to-use, computationally-efficient and easy-to-interpret solution. Interaction terms are important insights and sometimes understanding correlations and covariance between channels and baseline variables are key to be able to understand better how each variable interacts with one another.

Hope it helps!
Thanks,
Leo

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants