BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//AMSE//Event Calendar//FR
CALSCALE:GREGORIAN
METHOD:PUBLISH
BEGIN:VEVENT
UID:event-10123@amse-aixmarseille.fr
DTSTAMP:20260430T175748Z
CREATED:20260430T175748Z
LAST-MODIFIED:20260430T175748Z
STATUS:CONFIRMED
SEQUENCE:0
SUMMARY:big data and econometrics seminar - Sébastien Saurin
DTSTART:20230620T120000Z
DTEND:20230620T133000Z
DESCRIPTION:We introduce the XPER (eXplainable PERformance) methodology to 
 measure the specific contribution of the input features to the predictive o
 r economic performance of a model. Our methodology offers several advantage
 s. First\, it is both model-agnostic and performance metric-agnostic. Secon
 d\, XPER is theoretically founded as it is based on Shapley values. Third\,
  the interpretation of the benchmark\, which is inherent in any Shapley val
 ue decomposition\, is meaningful in our context. Fourth\, XPER is not plagu
 ed by model specification error\, as it does not require re-estimating the 
 model. Fifth\, it can be implemented either at the model level or at the in
 dividual level. In an application based on auto loans\, we find that perfor
 mance can be explained by a surprisingly small number of features. XPER dec
 ompositions are rather stable across metrics\, yet some feature contributio
 ns switch sign across metrics. Our analysis also shows that explaining mode
 l forecasts and model performance are two distinct tasks.\\n\\nContact: Mic
 hel Lubrano : michel.lubrano[at]univ-amu.frPierre Michel : pierre.michel[a
 t]univ-amu.fr\n\nPlus d'informations: https://amse-aixmarseille.fr/fr/evene
 ments/s%C3%A9bastien-saurin
LOCATION:Îlot Bernard du Bois - Salle 17\, AMU - AMSE\, 5-9 boulevard Maur
 ice Bourdet\, 13001 Marseille
URL;VALUE=URI:https://amse-aixmarseille.fr/fr/evenements/s%C3%A9bastien-saurin
CONTACT:Michel Lubrano : michel.lubrano[at]univ-amu.frPierre Michel :&nbsp\
 ;pierre.michel[at]univ-amu.fr
TRANSP:OPAQUE
END:VEVENT
END:VCALENDAR