From COVID-19 Testing to Election Prediction: How Small Are Our Big Data?: Professor Xiao-Li Meng (Harvard University)

Professor Meng discussed the term “Big Data” emphasizes data quantity, not quality. Professor Meng tackled important questions such as “What will be the effective sample size when we take into account the deterioration of data quality because of, for example, the selection bias in COVID-19 testing or the non-response bias in 2016 US Election polling results?” Professor Meng provided an answer to such questions, based on the concept developed in Meng (2018) Statistical paradises and paradoxes in big data (I): Law of large populations, bigdata paradox, and the 2016 US presidential election. Annals of Applied Statistics. Professor Meng also discussed briefly the application of this new concept for 2020 US Election, as reported in Isakov and Kuriwaki (2020) Towards Principled Unskewing: Viewing 2020 Election Polls Through a Corrective Lens from 2016. Harvard Data Science Review. More information can be found below.

- Professor Meng' article in Scientific American. This is written for the general public.
[ Ссылка ]

- Harvard Data Science Review (link is given below).
[ Ссылка ]

Format: 35 minutes lecture + 55 minutes discussion and Q&A

This is a recorded video (10 December 2020) from Virtual Seminar Series" Frontiers of Big Data, AI, and Analytics", co-organized by
Tomohiro Ando (Melbourne Business School, University of Melbourne)
Robert Kohn (UNSW Business School, University of New South Wales)
Valentin Zelenyuk (School of Economics, University of Queensland)

Our event series "Frontiers of Big Data, AI and Analytics" aims to unleash ideas and insights for harnessing the successful future of business & society. Therefore, the first part (lecture) focuses on cutting edge research/ideas and the second part (discussion) explores practical usage in business and society, bridging a gap between research/ideas and real business/managing organisations.

More information on this series is available at
[ Ссылка ]

свернуть