Morris Eaton

Given the current demographics, retirement policies, and investment practices of Americans, what can we expect the financial demand on social security to be in the year 2000? This is an example of the "statistical prediction problem": You have some data and a statistical model, and you must predict a quantity not yet observed. Over the past few years, much of my research has been directed at various forms of this problem. In some simple, low dimensional statistical problems, classical statistical procedures seem to predict fairly well. However, in more realistic situations when complex, high dimensional statistical models are often employed, there is now some evidence to suggest that current prediction methods are suspect. I am working on problems related to the effect of dimensionality on standard prediction methods. The ultimate goal of my research is to develop better predictive methods. Philosophically, my approach to the prediction problem is based on ideas associated with Bayesian methods and statistical decision theory. My previous research in multivariate statistical analysis has given me many of the mathematical tools necessary for my work in prediction. In particular, applications of group theory in statistics is a recurring theme in my publications. Much of my research has been supported by grants from the National Science Foundation.