BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//pyconde-pydata-2026//speaker//LHFSUL
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-pyconde-pydata-2026-TB9WYZ@pretalx.com
DTSTART;TZID=CET:20260416T140000
DTEND;TZID=CET:20260416T143000
DESCRIPTION:How do you evaluate performance when you predict more than 10 m
 illion time series each day? While a good plot can be worth more than a th
 ousand metrics for a single time series\, with large-scale machine learnin
 g models implemented with *LightGBM* and *PyTorch* we have to resort to me
 aningful aggregations. We will share insights and learnings from the past 
 2 years of deploying and operating our article-level demand forecasting mo
 dels at the pricing department of Zalando.\nThis talk moves beyond basic m
 etrics to showcase the pitfalls of aggregated error measures and the best 
 practices we’ve developed to keep our stakeholders informed and our mode
 ls accurate.
DTSTAMP:20260412T141726Z
LOCATION:Titanium [2nd Floor]
SUMMARY:How to compare apples with oranges: Proper evaluation of article-le
 vel demand forecasts - Stefan Birr\, Mones Raslan
URL:https://pretalx.com/pyconde-pydata-2026/talk/TB9WYZ/
END:VEVENT
END:VCALENDAR