Causal inference by using invariant prediction: identification and confidence intervals. With discussion and authors' reply
From MaRDI portal
Abstract: What is the difference of a prediction that is made with a causal model and a non-causal model? Suppose we intervene on the predictor variables or change the whole environment. The predictions from a causal model will in general work as well under interventions as for observational data. In contrast, predictions from a non-causal model can potentially be very wrong if we actively intervene on variables. Here, we propose to exploit this invariance of a prediction under a causal model for causal inference: given different experimental settings (for example various interventions) we collect all models that do show invariance in their predictive accuracy across settings and interventions. The causal model will be a member of this set of models with high probability. This approach yields valid confidence intervals for the causal relationships in quite general scenarios. We examine the example of structural equation models in more detail and provide sufficient assumptions under which the set of causal predictors becomes identifiable. We further investigate robustness properties of our approach under model misspecification and discuss possible extensions. The empirical properties are studied for various data sets, including large-scale gene perturbation experiments.
Recommendations
Cited in
(89)- Stabilizing variable selection and regression
- The Hardness of Conditional Independence Testing and the Generalised Covariance Measure
- Learning stable and predictive structures in kinetic systems
- Statistics for big data: a perspective
- Network modeling in biology: statistical methods for gene and brain networks
- Conditional variance penalties and domain shift robustness
- Towards out of distribution generalization for problems in mechanics
- Beyond the mean: a flexible framework for studying causal effects using linear models
- Regularized high dimension low tubal-rank tensor regression
- Models as approximations. II. A model-free theory of parametric regression
- Discussion: Models as approximations
- Rejoinder: Models as approximations
- Invariance, causality and robustness
- Comment: Invariance and causal inference
- Rejoinder: Invariance, causality and robustness
- Comments on ``Data science, big data and statistics
- Two-sample instrumental variable analyses using heterogeneous samples
- Causal Dantzig: fast inference in linear structural equation models with hidden variables under additive interventions
- scientific article; zbMATH DE number 6982327 (Why is no real title available?)
- Causal network reconstruction from time series: from theoretical assumptions to practical estimation
- Learning linear non-Gaussian causal models in the presence of latent variables
- Switching regression models and causal inference in the presence of discrete latent variables
- Causal discovery from heterogeneous/nonstationary data
- scientific article; zbMATH DE number 7255130 (Why is no real title available?)
- scientific article; zbMATH DE number 7626776 (Why is no real title available?)
- On semiparametric instrumental variable estimation of average treatment effects through data fusion
- Likelihood ratio tests for a large directed acyclic graph
- Reconstructing regime-dependent causal relationships from observational time series
- The blessings of multiple causes
- Causal learning via manifold regularization
- Robustifying independent component analysis by adjusting for group-wise stationary noise
- nonlinearICP
- Invariant causal prediction for sequential data
- Veridical data science
- Toward Causal Inference for Spatio-Temporal Data: Conflict and Forest Loss in Colombia
- tramicp
- Network Structure Learning Under Uncertain Interventions
- Deconfounding and Causal Regularisation for Stability and External Validity
- Estimation of Gaussian directed acyclic graphs using partial ordering information with applications to DREAM3 networks and dairy cattle data
- Learning to increase the power of conditional randomization tests
- Sound and complete causal identification with latent variables given local background knowledge
- Distributionally robust and generalizable inference
- Federated causal inference in heterogeneous observational data
- Conformal Sensitivity Analysis for Individual Treatment Effects
- Improved baselines for causal structure learning on interventional data
- Causal modelling of heavy-tailed variables and confounders with application to river flow
- Estimating causal effects with hidden confounding using instrumental variables and environments
- Online feature selection for multi-source streaming features
- From statistical to causal learning
- Higher-Order Least Squares: Assessing Partial Goodness of Fit of Linear Causal Models
- A Two-Sample Conditional Distribution Test Using Conformal Prediction and Weighted Rank Sum
- Individualized causal discovery with latent trajectory embedded Bayesian networks
- Data integration in causal inference
- Detecting responsible nodes in differential Bayesian networks
- Variable selection in linear regression models: choosing the best subset is not always the best choice
- Corrected score methods for estimating Bayesian networks with error-prone nodes
- Statistical Inference for Maximin Effects: Identifying Stable Associations across Multiple Studies
- Causal inference for oncology: past developments and current challenges
- A consistent estimator for confounding strength
- Environment invariant linear least squares
- Domain adaptation under hidden confounding
- Prominent roles of conditionally invariant components in domain adaptation: theory and algorithms
- Causality pursuit from heterogeneous environments via neural adversarial invariance learning
- Editorial: Special issue on ``Learning across distribution shifts
- Engression: extrapolation through the Lens of distributional regression
- Graphical criteria for the identification of marginal causal effects in continuous-time survival and event-history analyses
- Invariant probabilistic prediction
- Too many, too improbable: testing joint hypotheses and closed testing shortcuts
- Invariant subspace decomposition
- Model-Based Causal Feature Selection for General Response Types
- Domain generalization via content factors isolation: a two-level latent variable modeling approach
- De-biased two-sample U-statistics with application to conditional distribution testing
- Optimal transport and Wasserstein distances for causal models
- Toward Interpretable Deep Generative Models via Causal Representation Learning
- Confidence Sets for Causal Orderings
- Causality-Oriented Robustness: Exploiting General Noise Interventions
- Structural restrictions in local causal discovery: identifying direct causes of a target variable
- Structure learning and causal effect estimation from unbalanced groups
- Automated scientific discovery: from equation discovery to autonomous discovery systems
- Spectrally Deconfounded Random Forests
- Identifiability of causal graphs under non-additive conditionally parametric causal models
- Root cause discovery via permutations and Cholesky decomposition
- Learning When the Concept Shifts: Confounding, Invariance, and Dimension Reduction
- Neural causal regularization: extending causal invariance to deep models
- Bayesian graphs of intelligent causation
- Fast and Robust Invariant Generalized Linear Models
- Tests for differential Gaussian Bayesian networks based on quadratic inference functions
- Causal network learning with non-invertible functional relationships
- Marginal integration for nonparametric causal inference
This page was built for publication: Causal inference by using invariant prediction: identification and confidence intervals. With discussion and authors' reply
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q153568)
