Implicit Surveys at Facebook


MR2 - Centre for Mathematical Sciences
Cambridge, United Kingdom

Tuesday, 21 February 2017
from 16:00 to 17:00

Daniel Haimovich

Surveys might bring up images of random digit dialling and long questionnaires, but statistical surveys are being reimagined in the internet age. The typical “Big Data” dataset is generated when a group of people interact with some product. In some cases, we could think about this group of people as an implicit survey sample coming out of a larger population, and try to learn about it. I’ll describe some of the work we do, including traditional survey statistics methods which are as useful as they used to be and some new solutions to new problems.