BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//ox.ac.uk//NONSGML oxford.event//EN
X-WR-TIMEZONE:Europe/London
BEGIN:VTIMEZONE
TZID:Europe/London
X-LIC-LOCATION:Europe/London
BEGIN:DAYLIGHT
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:BST
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:GMT
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:Cross-temporal Forecast Reconciliation Using Machine Learning
DTSTART;TZID=Europe/London:20260605T141500
DTEND;TZID=Europe/London:20260605T153000
DTSTAMP:20260512T071951Z
UID:d497d98d-6219-f111-8342-7c1e522d9057
CREATED:20260306T140201Z
DESCRIPTION:Many forecasting tasks involve multiple\, interrelated time se
 ries that must satisfy linear aggregation constraints\, where the componen
 ts collectively sum to the total. Ensuring such coherence across all aggre
 gation levels is the goal of forecast reconciliation\, which is essential 
 for consistent and aligned decision-making. In cross-temporal frameworks\,
  the focus of this talk\, these aggregation constraints extend across both
  cross-sectional and temporal dimensions. Existing literature primarily re
 lies on linear reconciliation methods\, which adjust base forecasts throug
 h linear transformations within a least-squares framework to satisfy aggre
 gation constraints. In this work\, we move beyond this paradigm and introd
 uce a non-linear forecast reconciliation approach for cross-temporal frame
 works. Our method directly and automatically produces cross-temporal coher
 ent forecasts by leveraging popular machine learning techniques. We empiri
 cally validate our framework on large-scale streaming datasets from a lead
 ing European on-demand delivery platform and a bicycle-sharing system in N
 ew York City.
LAST-MODIFIED:20260424T145738Z
LOCATION:Manor Road Building - Seminar Room C\, Seminar Room C Manor Road 
 Building Manor Road Oxford Oxfordshire OX1 3UQ United Kingdom
SPEAKER:Ines Wilms
END:VEVENT
END:VCALENDAR