Today’s computing systems can be thought of as interventions in people’s work and daily lives. But what are the outcomes of these interventions, and how can we tune these systems for desired outcomes? In this project we are building methods to estimate the impact of changes to a product feature or a business decision before actually committing to it. These questions require causal inference methods; without an A/B test, patterns and correlations can lead us astray. For more on causal inference, refer to our tutorial on causal inference at the 2018 KDD conference.
We have used some of our latest research to build a software library, DoWhy, that provides a unified interface for causal inference methods and automatically tests their robustness to assumptions. Refer to the paper and the software library on Github for more details.