Robyn is a semi-automated Marketing Mix Modeling package that applies machine learning to measure marketing channel effectiveness and optimize budget allocation. It combines ridge regression, multi-objective evolutionary algorithms for hyperparameter tuning, time-series decomposition for trend analysis, and gradient-based optimization to calculate adstock rates and saturation curves across media channels.