Causal Inference in Big Data: A Selective Overview

Dr. Hyunseung Kang of the University of Wisconsin Department of Statistics will present on the following topic.

Abstract:
Many big data methods, especially machine learning methods, attempt to find the strongest association between a set of predictors and outcome. However, association does not imply causation and these methods alone cannot answer questions about causality, such as "Does smoking cause lung cancer?" or more generally, "Does X cause Y"?

The talk will provide a selective overview of causal inference, a field of statistics that aims to quantify the causal effect of a treatment, policy, or program on an outcome of interest. We will discuss three topics, with an emphasis on applications to big data settings: (1) inferring causal effects via matching, (2) inferring causal effects by using machine learning methods, and (3) understanding causal peer effects in large online social networks.

The food at this meetup is sponsored by American Family. The drinks/networking is sponsored by Hortonworks. The meetup itself is sponsored by Zendesk.