BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//IORA - Institute of Operations Research and Analytics - ECPv6.15.11//NONSGML v1.0//EN
CALSCALE:GREGORIAN
METHOD:PUBLISH
X-WR-CALNAME:IORA - Institute of Operations Research and Analytics
X-ORIGINAL-URL:https://iora.nus.edu.sg
X-WR-CALDESC:Events for IORA - Institute of Operations Research and Analytics
REFRESH-INTERVAL;VALUE=DURATION:PT1H
X-Robots-Tag:noindex
X-PUBLISHED-TTL:PT1H
BEGIN:VTIMEZONE
TZID:Asia/Singapore
BEGIN:STANDARD
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
TZNAME:+08
DTSTART:20210101T000000
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTART;TZID=Asia/Singapore:20221111T100000
DTEND;TZID=Asia/Singapore:20221111T113000
DTSTAMP:20260421T153423
CREATED:20221101T081611Z
LAST-MODIFIED:20221107T010932Z
UID:16407-1668160800-1668166200@iora.nus.edu.sg
SUMMARY:IORA Seminar Series - Long Zhao
DESCRIPTION:Long Zhao is an assistant professor in the Department of Analytics & Operations (DAO) at NUS Business School\, National University of Singapore. He received his Ph.D. in Decision Sciences from the McCombs School of Business at the University of Texas at Austin. Dr. Zhao’s research interests lie in data-driven decision-making. His research has been the finalist of INFORMS Data mining best paper competition and the finance section best student paper competition. \n  \n\n\n\nVenue \nHon Sui Sen Memorial Library – HSS 4-5\n\n\nLink to Register \n(Hybrid session)\nhttps://nus-sg.zoom.us/meeting/register/tZcsd–vqT4pGtaBggmJWaMKIuN6uTstODTJ\n\n\nTitle\nConstructing Quantiles via Forecast Errors: Theory and Empirical Evidence\n\n\nAbstract\nProbabilistic forecasts (such as quantiles) are essential inputs to decision-making in the face of uncertainty. However\, the most common type often comes in the form of point forecasts. As such\, it is necessary for the decision maker to construct uncertainty measures around the obtained point forecasts. One simple approach proposed in the literature suggests leveraging historical forecast errors to create quantile estimators around the given point forecast (referred to as the E2Q method). The sample quantile and normal approximation are two popular E2Q estimators. The former relies on the empirical distribution of the forecast errors while the latter treats the underlying distribution as if it were normal. Despite their prevalence\, the relative performances of the two estimators remain unknown. In this paper\, we find that the performance of a quantile estimator depends on its bias and variance. In particular\, higher variance always leads to worse performance. Furthermore\, unbiasedness is never optimal for a fixed variance and becomes less and less appealing as variance increases. Thus\, as an asymptotically unbiased estimator\, the sample quantile is appealing only when its variance is small. We confirm our theoretical findings using the M5 forecast competition data. Since this competition consists of both the “accuracy” (point) and “uncertainty” (quantile) tracks\, we also compare the E2Q method with other methods that directly forecast quantiles. We found that the E2Q method using the top point forecasts can outperform the top direct quantile forecasts. This empirical finding suggests that the E2Q method can be a promising alternative to forecasting quantiles directly.
URL:https://iora.nus.edu.sg/events/iora-seminar-series-long-zhao/
CATEGORIES:IORA Seminar Series
END:VEVENT
END:VCALENDAR