BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//AMSE//Event Calendar//FR
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:event-7841@www.amse-aixmarseille.fr
DTSTAMP:20260502T114006Z
CREATED:20260502T114006Z
LAST-MODIFIED:20260502T114006Z
STATUS:CONFIRMED
SEQUENCE:0
SUMMARY:phd seminar - Gilles Hacheme
DTSTART:20210119T100000Z
DTEND:20210119T104500Z
DESCRIPTION:Machine Learning (ML) models\, such as Random Forest and Boosti
 ng\, have shown their ability to get very good prediction results compared 
 to standard econometric approaches such as Linear regression.  Indeed\, ML
  models can approximate a very complex relationship between a set of explan
 atory variables and a dependent variable. That is capturing non-parametrica
 lly non-linearities and interactions efficiently. Nonetheless\, ML models a
 re seen as back box models while standard econometric models are better in 
 terms of interpretability. In this paper\, our goal is making ML models mor
 e interpretable by opening the black box to get a model at the same time pe
 rformant and interpretable. We suggest a method combining Generalized Addit
 ive Models (GAM) and a variable selection method (whether LASSO or Autometr
 ics). The GAM part can capture non-linearities and the selection method is 
 used to capture relevant interaction variables. Our simulations and applica
 tions show that this method can get very close results to the ones of stand
 ard ML models\, while being much more interpretable.\\n\\nContact: Anushka 
 Chawla : anushka.chawla[at]univ-amu.frKenza Elass : kenza.elass[at]univ-amu
 .frCarolina Ulloa Suarez : carolina.ulloa-suarez[at]univ-amu.fr\n\nPlus d'i
 nformations: https://www.amse-aixmarseille.fr/fr/evenements/gilles-hacheme-
 1
URL;VALUE=URI:https://www.amse-aixmarseille.fr/fr/evenements/gilles-hacheme-1
CONTACT:Anushka Chawla : anushka.chawla[at]univ-amu.frKenza Elass : kenza.e
 lass[at]univ-amu.frCarolina Ulloa Suarez : carolina.ulloa-suarez[at]univ-am
 u.fr
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR