Activity Number:
|
10
- Challenges and Breakthroughs in Analyzing Big Survey Data
|
Type:
|
Invited
|
Date/Time:
|
Sunday, July 28, 2019 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Government Statistics Section
|
Abstract #300377
|
|
Title:
|
Relationship Mining in Big Data from Surveys Using Penalization and the Bag-Of-Little-Bootstraps
|
Author(s):
|
Snigdhansu Chatterjee* and Benjamin E. Bagozzi and Ujjal Kumar Mukherjee and Xuetong Sun
|
Companies:
|
University of Minnesota and University of Delaware and University of Illinois and University of Minnesota
|
Keywords:
|
Data mining;
survey;
high dimensional;
rare event;
penalization;
bag-of-little bootstraps
|
Abstract:
|
We discuss high dimensional and large scale inference procedures under various survey distortions, systematic and non-ignorable non-responses and drop-outs, using a bag-of-little-bootstraps and penalization-based approach. We present a scheme for inference with biased survey data. Our methods may be used for studying how probabilities of rare events may be linked to explanatory variables even in biased data. We present several theoretical results, and illustrate the proposed methodology on data on political conflicts gathered from multiple federal agencies.
|
Authors who are presenting talks have a * after their name.