9321402 Schaffer The project is to learn about how statisticians would tackle the problems that serve as a central testbed for machine learning research (e.g., the UCI repositioning). In particular, the proposed work will be concerned with the following four questions: (1) What model classes and fitting procedures do statisticians choose when faced with supervised learning problems drawn from the UCI repository? (2) On what basis do they make their choices? (3) How are results of data analysis interpreted to make predictions? (4) How accurate are these predictions?
