PHYSTAT informal review: Discrete Profiling

by Anthony C. Davison (EPFL - EPF Lausanne), Matthew William Kenzie (University of Cambridge (GB))

Wednesday 4 Dec 2024, 15:30 → 17:00 Europe/Zurich

This is the a PHYSTAT Informal Review event. In this virtual format, a Tandem consisting of a physicist and a statistician will review a statistical method introduced by one of the parties or a general critical analysis topic from the Physicist's and statistician's perspectives. The virtual events comprise: two 20+10 min. complementary presentations followed by ~30 minutes of general discussion.

Today Matt Kenzie (physicists, Cambridge) together with Anthony Davison (statistician, EPFL) will review the topic "Discrete Profiling".

Agenda:

• 3.30 pm Opening 
• 3.30 pm Presentation Matt Kenzie 20'+10')
• 4 pm Presentation Anthony Davison (20'+10')
• 4.30 pm General Discussion and Closing (30')

 

Abstract: 

A common problem in data analysis is that the functional form (and not just the parameter values) of the underlying model describing the data is unknown a priori. In many such cases an additional uncertainty should be assigned to the parameters of interest to account for the lack of knowledge of the underlying model. The "discrete profiling" or “envelope” method attempts to assign the appropriate uncertainty to the parameter of interest by profiling over the choice of possible models.  Discrete profiling requires the user to choose the models to be profiled over and to choose penalties for models with different numbers of free parameters.  A statistical formulation of the approach sheds some light on these issues.

Anthony_Davison.pdf

arXiv:1408.6865

GMT20241204-143121_Recording_2240x1260.mp4

mkenzie_experimental_phystat_envelope.pdf