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.
feat: Exposure fitting is being reactivated. When paid_media_vars are different than paid_media_spend, paid_media_vars will be used for fitting. The extrapolation from exposure to spend uses the metric CPE (cost per exposure) = spend / exposure. Respectively, Michaelis Menten function for spend-exposure fitting is deprecated, because it doesn't provide significant improvement in fitting, while being very error-prone. The missing nonlinear relationship between spend and exposure will be address by curve calibration.
feat: curve calibration. The new robyn_calibrate feature works standalone and without the other main Robyn functions. It consumes a dataframe a two columns: spend and response. This allows users to provide any saturation curves as source of truth to improve Robyn's saturation parameter identification and narrow down sample the ranges for alpha and gamma. Typical source of truth for saturation curves are Meta conversion lift, as well as Halo cumulative reach.
feat: reach and frequency allocator allows users to find optimum reach and frequency combination for media planning.
Refactor: aligned and simplified nonlinear transformation process across functions like run_transformation, robyn_response, robyn_pareto and robyn_allocator
Refactor: inflexion point calculation now changed from .dot_product(range(x), gamma) to sum(x) * gamma to allow more flexibility in curve estimation
Type of change
Unit test (tbc)