BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:The data-driven (s\,S) policy: why you can have confidence in cens
 ored demand data - Gah-Yi Vahn (London School of Business)
DTSTART:20160129T160000Z
DTEND:20160129T170000Z
UID:TALK63640@talks.cam.ac.uk
CONTACT:Quentin Berthet
DESCRIPTION:I revisit the classical dynamic inventory management problem o
 f Scarf (1959) from a distribution-free\, data-driven perspective. I propo
 se a nonparametric estimation procedure for the optimal (s\, S) policy tha
 t yields an asymptotically optimal estimated policy and analytically deriv
 e confidence intervals around this policy. I also derive a confidence boun
 d on the estimated total cost\, which\, in the case of zero setup cost\, i
 nterestingly is directly proportional to the size of the confidence interv
 als of the estimated policy. I further consider having a portion of the da
 ta censored from past ordering decisions. I show that the intuitive proced
 ure of correcting for censoring in the demand data directly yields an inco
 nsistent estimate of the optimal policy. I then show how to correctly use 
 the censored data to obtain consistent decisions and derive confidence int
 ervals for this policy. Remarkably\, under some conditions\, ordering deci
 sions estimated with partially censored data may be more precise than with
  fully uncensored data\, and there exists an optimal amount of censored da
 ta to minimise the mean square error (MSE). 
LOCATION:MR12\, Centre for Mathematical Sciences\, Wilberforce Road\, Camb
 ridge.
END:VEVENT
END:VCALENDAR