crumble

Flexible and General Mediation Analysis Using Riesz Representers

Implements a modern, unified estimation strategy for common mediation estimands (natural effects, organic effects, interventional effects, and recanting twins) in combination with modified treatment policies as described in Liu, Williams, Rudolph, and Díaz (2024) <doi:10.48550/arXiv.2408.14620>. Estimation makes use of recent advancements in Riesz-learning to estimate a set of required nuisance parameters with deep learning. The result is the capability to estimate mediation effects with binary, categorical, continuous, or multivariate exposures with high-dimensional mediators and mediator-outcome confounders using machine learning.

Versions across snapshots

Dependencies (latest)