RealCause: Realistic Causal Inference Benchmarking
arXiv:2011.15007
Abstract
There are many different causal effect estimators in causal inference. However, it is unclear how to choose between these estimators because there is no ground-truth for causal effects. A commonly used option is to simulate synthetic data, where the ground-truth is known. However, the best causal estimators on synthetic data are unlikely to be the best causal estimators on real data. An ideal benchmark for causal estimators would both (a) yield ground-truth values of the causal effects and (b) be representative of real data. Using flexible generative models, we provide a benchmark that both yields ground-truth and is realistic. Using this benchmark, we evaluate over 1500 different causal estimators and provide evidence that it is rational to choose hyperparameters for causal estimators using predictive metrics.
References in corpus (8)
- Scikit-learn: Machine Learning in Python
- Meta-learners for Estimating Heterogeneous Treatment Effects using Machine Learning
- Causal Effect Inference with Deep Latent-Variable Models
- Removing Hidden Confounding by Experimental Grounding
- An Evaluation Toolkit to Guide Model Selection and Cohort Definition in Causal Inference
- Benchmarking Framework for Performance-Evaluation of Causal Inference Analysis
- Synth-Validation: Selecting the Best Causal Inference Method for a Given Dataset
- How well does your sampler really work?
Cited by in corpus (5)
- DoWhy: Addressing Challenges in Expressing and Validating Causal Assumptions
- Causal Effect Inference for Structured Treatments
- Generating Synthetic Text Data to Evaluate Causal Inference Methods
- Doing Great at Estimating CATE? On the Neglected Assumptions in Benchmark Comparisons of Treatment Effect Estimators
- ADCB: An Alzheimer's disease benchmark for evaluating observational estimators of causal effects