BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//CERN//INDICO//EN
BEGIN:VEVENT
SUMMARY:A generalized method for Sparse Partial Least Squares (Dual-SPLS):
theory and applications
DTSTART;VALUE=DATE-TIME:20210503T083000Z
DTEND;VALUE=DATE-TIME:20210503T093000Z
DTSTAMP;VALUE=DATE-TIME:20230605T154900Z
UID:indico-event-6595@indico.math.cnrs.fr
DESCRIPTION:Speakers: Louna Al Souki\n\nIn data analysis\, high dimensiona
lity is often a delicate obstacle to overcome. The problem can be solved b
y representing the data in a lower dimensional space using projection meth
ods like the Partial Least Squares regression (PLS) or by resorting to var
iable selection methods like the lasso approach. The Sparse Partial Least
Squares (SPLS) combines the two approaches in order to better interpret th
e results due to the sparsity imposed on the new directions. Several imple
mentations have been proposed. However\, problems of accuracy of predictio
ns and correct interpretation of regression coefficients arise in these ap
proaches. Hence we developed the Dual Sparse Partial Least Squares\, a fle
xible method that results in more accurate predictions and better interpre
tation of the coefficients due to their sparsity. In this paper we present
the theory behind Dual-SPLS and some applicative results on petroleum dat
a sets.\n\nhttps://indico.math.cnrs.fr/event/6595/
LOCATION: BigBlueButton Platform (Webinar)
URL:https://indico.math.cnrs.fr/event/6595/
END:VEVENT
END:VCALENDAR