Machine Learning

Counterfactual explainability of black-box prediction models

We propose a counterfactual notion of explanability for black-box prediction models.

Comment: Will competition-winning methods for causal inference also succeed in practice?

This is an invited commentary for Statistical Science on the causal inference data competition in ACIC 2016.

Causal interpretations of black-box models

We link Friedman's partial dependence plot with Pearl's backdoor adjustment formula. We discuss situations when possible causal interpretations can be made for black-box machine learning models.