Dr. James Livsey, Principal Researcher, United States Census Bureau's Center for Statistical Research and Methodology
Statistics Seminar Series (Hybrid)
Signal Extraction with Latent Component Models
As the US Census Bureau aims to increase the frequency of data dissemination, the topic of signal extraction has gained significant attention. This talk will delve into multiple research threads related to the challenges of performing signal extraction. Specifically, we introduce a class of difference stationary latent component models that can be used to identify structural components. Within this class, we will explore two open areas of research: multivariate extraction and fractional periodicities. These areas each present unique challenges. To illustrate the practical applications of this methodology, we will present case studies involving daily immigration and weekly business formation statistics.
Dr. James Livsey is a Principal Researcher and Mathematical Statistician in the Time Series Group at the United States Census Bureau's Center for Statistical Research and Methodology. He also holds an adjunct faculty position at the University of Buffalo, where he teaches classes for the Mathematics Department and in the Institute for Artificial Intelligence and Data Science. He holds a Ph.d. from Clemson University and works on many innovative statistical methodologies for a wide range of applications, including seasonal adjustment, signal extraction, and spatial modeling.