BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Randomisation for Agnostic Selective Inference - Alastair Young (I
 mperial College\, London)
DTSTART:20230120T140000Z
DTEND:20230120T150000Z
UID:TALK194890@talks.cam.ac.uk
CONTACT:Qingyuan Zhao
DESCRIPTION:In contemporary statistical applications\, selection of the fo
 rmal inferential problem is typically done after some level of interaction
  with the data. Usually\, an initial exploratory analysis is used to ident
 ify those aspects of some population that appear interesting\, and then th
 e same dataset is used to learn about them. Such double use of the data in
 validates classical inferential procedures: selective inference aims to co
 rrect for `data snooping' and provide valid inference on selected paramete
 rs. We propose a powerful procedure for selective inference in selection-a
 gnostic settings\, where the selection mechanism is unknown or difficult t
 o handle analytically.\n\nOur method operates by performing the explorator
 y analysis on an artificially randomised version of the data and basing th
 e inference on an orthogonal complement\, which is independent of selectio
 n by construction in Gaussian settings. We discuss the relationship of the
  method with data splitting and describe how\, under mild conditions\, the
  method is asymptotically justified in the non-Gaussian context.\n \nThis 
 is joint work with Daniel Garcia Rasines.
LOCATION:MR12\, Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR