BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Parameterizing and Simulating from Causal Models
DTSTART:20230213T143000
DTEND:20230213T153000
DTSTAMP:20260407T014827Z
UID:980b4b2943854a14c7d608b205a7726612f596218fa2df8bb613f2d8
CATEGORIES:Conferences - Seminars
DESCRIPTION:Robin EVANS\, University of Oxford\nSeminar in Mathematics\nAb
 stract: Many statistical problems in causal inference involve a probabilit
 y distribution other than the one from which data are actually observed\
 ; as an additional complication\, the object of interest is often a margi
 nal quantity of this other probability distribution. This creates many p
 ractical complications for statistical inference\, even where the proble
 m is non-parametrically identified. In particular\, it is difficult to pe
 rform likelihood-based inference\, or even to simulate from the model in
  a general way. \n\nWe introduce the frugal parameterization\, which pla
 ces the causal effect of interest at its centre\, and then builds the res
 t of the model around it. We do this in a way that provides a recipe for 
 constructing a regular\, non-redundant parameterization using causal quan
 tities of interest. In the case of discrete variables we can use odds ra
 tios to complete the parameterization\, while in the continuous case copu
 las are the natural choice. \n\nOur methods allow us to construct and sim
 ulate from models with parametrically specified
LOCATION:MA B1 11 https://plan.epfl.ch/?room==MA%20B1%2011 https://epfl.zo
 om.us/j/69915839681
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR