Using hierarchical models to figure out the relationship between physical and mental disabilities in children and various reported symptoms
Building a large and representative dataset of fake news collected from the internet. The work involved thinking through how to define "fake news" and how to operationalize the concept so that news items can be classified as "fake" or "real" using crowdsourcing.
Predicting the success of first-generation and low-income college students using machine learning models, and interpreting the complex machine learning models to draw conclusions about interventions that may be effective
Predicting stock returns from transcripts of earning calls using machine learning.
Exploring the relationship between eviction filing fees and eviction rates using hierarchical models
Building a face classification system for determining whether a person looks like a model; exploring the fairness of the system.