Marginal integration for nonparametric causal inference
arXiv:1405.1868 · doi:10.1214/15-EJS1075
Abstract
We consider the problem of inferring the total causal effect of a single variable intervention on a (response) variable of interest. We propose a certain marginal integration regression technique for a very general class of potentially nonlinear structural equation models (SEMs) with known structure, or at least known superset of adjustment variables: we call the procedure S-mint regression. We easily derive that it achieves the convergence rate as for nonparametric regression: for example, single variable intervention effects can be estimated with convergence rate assuming smoothness with twice differentiable functions. Our result can also be seen as a major robustness property with respect to model misspecification which goes much beyond the notion of double robustness. Furthermore, when the structure of the SEM is not known, we can estimate (the equivalence class of) the directed acyclic graph corresponding to the SEM, and then proceed by using S-mint based on these estimates. We empirically compare the S-mint regression method with more classical approaches and argue that the former is indeed more robust, more reliable and substantially simpler.
arXiv comment40 pages, 14 figures
References in corpus (13)
- Boosting Algorithms: Regularization, Prediction and Model Fitting
- Estimating high-dimensional directed acyclic graphs with the PC-algorithm
- Estimating high-dimensional intervention effects from observational data
- Learning high-dimensional directed acyclic graphs with latent and selection variables
- Characterization and Greedy Learning of Interventional Markov Equivalence Classes of Directed Acyclic Graphs
- CAM: Causal additive models, high-dimensional order search and penalized regression
- Identifiability of Gaussian structural equation models with equal error variances
- Ordering-Based Search: A Simple and Effective Algorithm for Learning Bayesian Networks
- -penalized maximum likelihood for sparse directed acyclic graphs
- High-dimensional learning of linear causal networks via inverse covariance estimation
- Two Optimal Strategies for Active Learning of Causal Models from Interventional Data
- Jointly interventional and observational data: estimation of interventional Markov equivalence classes of directed acyclic graphs
- Score-based Causal Learning in Additive Noise Models