Faculty and students from neighboring universities are cordially invited. Please contact Jamie Stafford for more information.
I propose a new criterion for model selection in prediction problems. The covariance inflation criterion adjusts the training error by the average covariance of the predictions and responses, when the prediction rule is applied to permuted versions of the data set. This criterion can be applied to general prediction problems (for example regression or classification), and to general prediction rules (for example stepwise regression, tree-based models and neural nets). As a byproduct we obtain a measure of the effective number of parameters used by an adaptive procedure. I relate the covariance inflation criterion to other model selection procedures and illustrate its use in some regression and classification problems. I also revisit the conditional bootstrap approach to model selection.
Brad will discuss an analysis of astrophysical data for Quasars. The variables available are redshift and luminosity where the latter is truncated. The two questions of interest are whether the two variables are independent and if so what is the luminosity distribution? It is surprising that the methods used also have applications in biostatistics.