Paper: Doubly robust treatment effect estimation with missing attributes

New solutions for causal inference with missing values.

We propose two consistent doubly robust treatment effect estimators handling missing attributes (missing values in the covariates). These estimators are based on recent results for likelihood-based methods and tree-based methods handling missing values.

For a first version of our paper click here. (First update: 2019-06-25, last update: 2019-10-23.)

And a poster presented at the Data Science Summer School 2019.

Imke Mayer
PhD Student in Statistics and Applied Mathematics