** ANNOUNCEMENT **
Command, Control and Interoperability Center for Advanced Data Analysis – CCICADA Research Group Lab Meeting
Date/Time: Thursday, May 25, 2017/12:15pm to 1:00pm
Computing Research & Education Building (CoRE), Busch Campus, Rutgers, the State University of NJ
Location to be determined (either 4th Floor Conference Room – 433 or 6th Floor Conference Room 631 – additional information will be forwarded)
Ilya Muchnik, DIMACS at Rutgers University and LongShortWay, Inc.
Clustering to Build Optimal Stratified Sampling
This talk will bring attention on an area called in social statistics as “optimal stratification sampling”. It is a group of mathematical and engineering methods to build a random set of observations under a given cost limitation, which guarantees an optimal representation for unknown population, on which one has to analyze some social/economical events.
You can look on the technology as complementary to the deep learning, which tries to collect for an analysis on a random set as large as possible with hope, that a larger set should be more representative. In general it is not true. The collection can become biased. To collect a data with a good representation one has to follow articular requests.