|
Abstract:
|
The immediate need for fast and effective data mining for massive data gives rise to a recent new field in statistics: large-scale multiple comparisons, where thousands or even millions of hypotheses are testing simultaneously. In this talk, we consider a problem of particular interest in this field: estimating the proportion of non-null effects. We report some recent progress on estimating the proportion, ranging from the non-sparse regime to the very sparse regime. In particular, we report a method for constructing oracle equivalence of the proportion, which equals to the proportion for any underlying parameters. The oracle naturally yields real estimators, which are shown to be uniformly consistent for the proportion over a wide class of situations. This talk is based joint works with (alphabetically) Tony Cai, David Donoho, Mark Low, Jie Peng, and Pei Wang.
|
