Online Program Home
My Program

Abstract Details

Activity Number: 138 - Modeling Applications for Backcasting, Nowcasting and Forecasting
Type: Contributed
Date/Time: Monday, July 29, 2019 : 8:30 AM to 10:20 AM
Sponsor: Survey Research Methods Section
Abstract #304683 Presentation
Title: Consideration of Unsupervised Learning in the Detection of Systemic Errors Within the Current Employment Statistics Survey
Author(s): Matthew Corrigan*
Companies: Bureau of Labor Statistics
Keywords: Establishment Survey; Total Survey Error; Mixture Modeling; flexmix

This paper will model the closed form elements from the decomposition of sampling errors. Errors may be accurately decomposed into elements of total survey error several months after the Current Employment Statistics (CES) survey results are published using the census from the Quarterly Census of Earnings and Wages (QCEW) administrative file. We have attempted to use unsupervised learning with the idea that if the errors cluster consistently across time, we can identify the micro-data associated with the systemic error and perhaps remedy the causes for such a consistent error reducing the overall error affecting CES estimates. The purpose of this study is to evaluate if the decomposed errors can be clustered in a way that gives us insight into where to look at the root causes of their similarity within various levels of industry or geographic regions.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2019 program