Biostatistics Weekly Seminar

A topic from probability theory, the Tower Property, will be used to present two novel methods in applied statistics. The first involves the perennial problem of missing values in one's dataset, in contexts in which prediction, rather than effect estimation, is the prime focus. The second deals with the topic of fairness in machine learning, currently of major ethical concerns. (No background in machine learning will be needed to follow the talk.) The question at hand is, How can we predict some variable Y, using a linear/generalized linear model or a nonparametric method, from variables X without being biased in terms of a sensitive variable (e.g. race or gender) S?

Hybrid: 2525 West End Ave, Room 11105 // Zoom link below
12 June 2024
1:30pm