Short Course: Text-As-Data Methods with Brandon Stewart

TRIADS is pleased to announce its first short course, being offered in partnership with the Department of Political Science.

Brandon Stewart (Princeton University Sociology) will be teaching a six-day short course on text-as-data methods for the social sciences. The class will cover a broad range of topics, including supervised learning, topic models, text annotation, large language models, and more. The course will also discuss issues related to using advanced machine learning methods for social science research.

Dates: The class will be held in person on August 14, 15, 16 and 22, 23, 24.

Location: The class will take place in person in Seigle Hall on the Danforth Campus.  Recordings of the session will be made available to participants for internal Washington University use only.

Registration: This class is not for credit, but is open to all Ph.D. students, postdocs, and faculty at Washington University. While we are offering the course at no cost, all participants are expected to be actively engaged in the course, and complete assigned readings and problem sets. If you are interested in taking this class, you must email Dahjin (Jin) Kim at dahjin.kim@wustl.edu.

Course Format: The course will consist of morning in-class lecture sessions. Afternoons will be in-person lab sessions to work on problem sets in groups. A larger problem set may be assigned on the 16 and due on the 22. TAs will be available to help with assignments.

Pre-requisites: The class will use linear regression as a springboard for introducing more complex ideas. Students should have completed a graduate-level linear models class, preferably one that relies on concepts from linear algebra. All code examples will be present in R, but almost everything will also be possible to complete in Python. Students must be able to work in one of these languages to take the course.