BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:Linear Models for Distributional Data Analysis
DTSTART:20241122T100000Z
DTEND:20241122T104000Z
DTSTAMP:20260617T140000Z
UID:indico-event-13451@indico.math.cnrs.fr
DESCRIPTION:Speakers: Paula Brito\n\nIn  classical Statistics and Multiv
 ariate Data Analysis data is usually represented in a matrix where each ro
 w represents a statistical unit\, or “individual”\, for which one sing
 le value is recorded for each numerical or categorical variable (in column
 s). This representation model is however too restricted when the data to b
 e analysed comprises variability. That is the case when the entities under
  analysis are not single elements\, but groups formed on the basis of some
  given common properties and the observed variability within each group sh
 ould be taken into account. To this aim\, new variable types have been int
 roduced\, whose realizations are not single real values or categories\, bu
 t sets\, intervals\, or distributions over a given domain.  Symbolic Dat
 a Analysis provides a framework for the representation and analysis of suc
 h data\, taking into account their inherent variability.  In this talk\,
  we consider the case of aggregate numerical data described by empirical d
 istributions\, known as histogram data. Linear models for such distributio
 nal variables are proposed\, which rely on the representation of histogram
 s or intervals by the associated quantile functions\, under specific assum
 ptions. These then allow for multivariate analysis of distributional-value
 d data\, e.g. multiple linear regression or linear discriminant analysis. 
 Applications of the proposed approach will be presented. \n\nhttps://indi
 co.math.cnrs.fr/event/13451/
LOCATION:Salle de conférence (XLIM)
URL:https://indico.math.cnrs.fr/event/13451/
END:VEVENT
END:VCALENDAR