Forecasting expected_spend_days

facebookexperimental / Robyn

Robyn is an experimental, AI/ML-powered and open sourced Marketing Mix Modeling (MMM) package from Meta Marketing Science. Our mission is to democratise modeling knowledge, inspire the industry through innovation, reduce human bias in the modeling process & build a strong open source marketing science community.

https://facebookexperimental.github.io/Robyn/

MIT License

1.07k stars 322 forks source link

Forecasting expected_spend_days #940

Open KryeKuzhinieri opened 3 months ago

KryeKuzhinieri commented 3 months ago

In the previous versions of Robyn we had max_response_expected_spend with expected_spend_days and expected_spend for robyn_allocator. Now in the new version this seems to have been changed to max_historical_response which is cool but the questions are:

How to predict the future, say, next 4 weeks (28 days)?
If there is no predicting the future, then what is the point of allocating data for past periods since we are doing calculations on the total budget?

gufengzhou commented 2 months ago

Sorry for the late reply. The latest version has allocator scenario of "max_response" and "target_efficiency". See ?robyn_allocator for more description.

Robyn doesn't offer future forecasting, so you're right that the allocator is doing past simulation. But to be honest the only difference between forecast and simulation is the forecast of trend/ season etc. For media variables, the optimised budget is derived exactly the same as the past simulation, based on the same saturation curve and same coefficient. And of course the trend forecast is a task by itself anyways.

We're aware that many have asked for the forecast feature. We'll consider it in the future.